Director of Research, Agentic AI
for section headings. Use - for any list of duties/requirements/benefits, and
for prose. Do NOT repeat the job title as a heading (the page already shows it) and do NOT put Job ID / requisition numbers as headings — fold minor metadata into a short
or drop it. Preserve ALL substantive information; do NOT invent or summarize away detail. Remove cookie/navigation/legal/EEO/equal-opportunity/scam-warning boilerplate, duplicated lines, and application instructions. Important: include a section ONLY when the posting has real, specific content for it — never emit a heading followed by filler/placeholder text (e.g. "commensurate with experience", "not specified", "N/A", "varies", "TBD") and never create Pay/Schedule/Benefits sections unless the posting actually states them. Prefer a few substantive sections over many thin ones.
We need to decide which sections to include. The posting has:
- Intro paragraph(s) about company, mission, role description. Could be "About the role" or maybe "Overview". But we need to use headings like About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. The instruction says use
for section headings (e.g. About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule). So we should map content to those headings where appropriate.
We have:
- About the role: could be the introductory description of the role and mission.
- Responsibilities: "What You'll Do" bullet list.
- Requirements/Qualifications: "What You'll Add To DigitalOcean" (required qualifications) and "Preferred Qualifications". Could combine into Requirements and Preferred Qualifications sections.
- Skills: maybe not separate; could be part of qualifications.
- Benefits: "Why You’ll Like Working for DigitalOcean" includes benefits, career development, well-being, rewards. Also compensation range is separate.
- Pay: Compensation Range $249,600 - $312,000.
- Schedule: hybrid role mentioned: "This is a hybrid role". Also maybe location? Not given. So we can include Schedule section with hybrid.
We must not create sections with filler. So we need to ensure each section has substantive content.
We need to output only HTML fragment, no /, no markdown fences.
We need to use - for any list of duties/requirements/benefits. So for responsibilities list, we need
- . For qualifications list also
- . For benefits also
- . For maybe skills? Not needed.
We need to preserve all substantive information; do not invent or summarize away detail. So we need to keep the exact wording? Probably we can paraphrase but must keep substance. Safer to keep the sentences as is, maybe break into paragraphs.
We need to remove boilerplate: EEO, equal-opportunity, scam-warning, application instructions, duplicated lines, cookie/navigation/legal. At end there is "DigitalOcean is an equal-opportunity employer..." and "Application Limit: You may apply to a maximum of 3 positions...". Those should be removed.
Also "JR: 2026-8125" is job ID; we should fold minor metadata into a short
or drop it. Probably drop it.
Also "Compensation Range $249,600 - $312,000This is a hybrid role" appears as a line without space. We need to separate.
Let's extract content.
Original text:
"Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world. DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior. We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean. This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders."
Then "What You'll Do" with bullet points.
Then "What You'll Add To DigitalOcean" with bullet points (required qualifications).
Then "Preferred Qualifications" with bullet points.
Then "Compensation Range $249,600 - $312,000This is a hybrid role JR: 2026-8125"
Then "Why You’ll Like Working for DigitalOcean" with paragraphs.
We need to produce sections:
- About the role: maybe combine the introductory paragraphs up to "This is a hybrid role by design:"? But we also have hybrid role mention later. Could put About the role as the intro describing mission and role.
- Responsibilities: the "What You'll Do" bullet list.
- Requirements: the "What You'll Add To DigitalOcean" bullet list (required qualifications). Could also include "Preferred Qualifications" as separate section or under Requirements? The instruction says prefer few substantive sections over many thin ones. Could combine Required and Preferred into one "Qualifications" section with subheadings? But we only allowed headings list: About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. So we could have Requirements for required qualifications, Qualifications for preferred qualifications, Skills maybe not needed. Or we could put both under Requirements and skip Qualifications. But we need to preserve all info; we can have both sections.
- Benefits: the "Why You’ll Like Working for DigitalOcean" paragraphs.
- Pay: compensation range.
- Schedule: hybrid role.
We need to ensure we don't create sections with filler. All have content.
Now format.
We need to use
for each section heading.
Inside each section, we need to use
for prose, and
- for any list.
Thus:
About the role
...
But the intro is multiple paragraphs; we can combine into one or multiple
. Probably multiple
okay.
We need to preserve all substantive information; we can split into paragraphs logically.
Let's craft About the role content:
First paragraph: "Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world."
Second paragraph: "DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior."
Third paragraph: "We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean."
Fourth paragraph: "This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders."
That covers the intro.
Now Responsibilities: "What You'll Do" bullet list.
We need to capture each bullet exactly.
From text:
"What You'll Do Set the research agenda Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and policies).Track emerging model architectures and specialized, domain-tuned model approaches, and translate what's relevant into DigitalOcean's product roadmap.Keep the agenda tightly coupled to product outcomes — every research bet should map to a measurable improvement in model selection, agent reliability, cost/latency, or task-success rate, not research for its own sake. Build and lead the team Grow the Applied Research team, hiring and mentoring research scientists and engineers.Establish the team's operating rhythm: how research questions get scoped, how experiments get run and evaluated, and how findings hand off to production teams.Represent Applied Research in cross-functional planning cycles alongside other engineering and product leaders, and make the case for headcount and investment on its own merits. Bridge research and product Partner directly with Inference Engine and the other platform and product engineering teams that own agent runtime and evaluation infrastructure to turn research into shipped capability.Turn research prototypes into production-ready capabilities in partnership with engineering — shipping research, not just publishing it.Communicate research trade-offs clearly to non-research stakeholders, including when a promising direction isn't ready for product investment yet. Represent DigitalOcean externally Build DigitalOcean's credibility in the applied agentic-AI research community through publications, talks, open-source contributions, or collaborations — where they serve product and hiring goals."
We need to split into bullet points. The original seems to have sections: Set the research agenda, Build and lead the team, Bridge research and product, Represent DigitalOcean externally. Each with sub-bullets.
We can create
- for each top-level bullet, and inside each li we can have
or just text. But instruction: use
- for any list of duties/requirements/benefits, and
for prose. So we can have
- containing prose (maybe
inside). Simpler: each li contains the text.
Thus:
Responsibilities
- Set the research agenda: Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and
for prose. Do NOT repeat the job title as a heading (the page already shows it) and do NOT put Job ID / requisition numbers as headings — fold minor metadata into a short
or drop it. Preserve ALL substantive information; do NOT invent or summarize away detail. Remove cookie/navigation/legal/EEO/equal-opportunity/scam-warning boilerplate, duplicated lines, and application instructions. Important: include a section ONLY when the posting has real, specific content for it — never emit a heading followed by filler/placeholder text (e.g. "commensurate with experience", "not specified", "N/A", "varies", "TBD") and never create Pay/Schedule/Benefits sections unless the posting actually states them. Prefer a few substantive sections over many thin ones. We need to decide which sections to include. The posting has: - Intro paragraph(s) about company, mission, role description. Could be "About the role" or maybe "Overview". But we need to use headings like About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. The instruction says use
for section headings (e.g. About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule). So we should map content to those headings where appropriate.
We have:
- About the role: could be the introductory description of the role and mission.
- Responsibilities: "What You'll Do" bullet list.
- Requirements/Qualifications: "What You'll Add To DigitalOcean" (required qualifications) and "Preferred Qualifications". Could combine into Requirements and Preferred Qualifications sections.
- Skills: maybe not separate; could be part of qualifications.
- Benefits: "Why You’ll Like Working for DigitalOcean" includes benefits, career development, well-being, rewards. Also compensation range is separate.
- Pay: Compensation Range $249,600 - $312,000.
- Schedule: hybrid role mentioned: "This is a hybrid role". Also maybe location? Not given. So we can include Schedule section with hybrid.
We must not create sections with filler. So we need to ensure each section has substantive content.
We need to output only HTML fragment, no /, no markdown fences.
We need to use - for any list of duties/requirements/benefits. So for responsibilities list, we need
- . For qualifications list also
- . For benefits also
- . For maybe skills? Not needed.
We need to preserve all substantive information; do not invent or summarize away detail. So we need to keep the exact wording? Probably we can paraphrase but must keep substance. Safer to keep the sentences as is, maybe break into paragraphs.
We need to remove boilerplate: EEO, equal-opportunity, scam-warning, application instructions, duplicated lines, cookie/navigation/legal. At end there is "DigitalOcean is an equal-opportunity employer..." and "Application Limit: You may apply to a maximum of 3 positions...". Those should be removed.
Also "JR: 2026-8125" is job ID; we should fold minor metadata into a short
or drop it. Probably drop it.
Also "Compensation Range $249,600 - $312,000This is a hybrid role" appears as a line without space. We need to separate.
Let's extract content.
Original text:
"Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world. DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior. We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean. This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders."
Then "What You'll Do" with bullet points.
Then "What You'll Add To DigitalOcean" with bullet points (required qualifications).
Then "Preferred Qualifications" with bullet points.
Then "Compensation Range $249,600 - $312,000This is a hybrid role JR: 2026-8125"
Then "Why You’ll Like Working for DigitalOcean" with paragraphs.
We need to produce sections:
- About the role: maybe combine the introductory paragraphs up to "This is a hybrid role by design:"? But we also have hybrid role mention later. Could put About the role as the intro describing mission and role.
- Responsibilities: the "What You'll Do" bullet list.
- Requirements: the "What You'll Add To DigitalOcean" bullet list (required qualifications). Could also include "Preferred Qualifications" as separate section or under Requirements? The instruction says prefer few substantive sections over many thin ones. Could combine Required and Preferred into one "Qualifications" section with subheadings? But we only allowed headings list: About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. So we could have Requirements for required qualifications, Qualifications for preferred qualifications, Skills maybe not needed. Or we could put both under Requirements and skip Qualifications. But we need to preserve all info; we can have both sections.
- Benefits: the "Why You’ll Like Working for DigitalOcean" paragraphs.
- Pay: compensation range.
- Schedule: hybrid role.
We need to ensure we don't create sections with filler. All have content.
Now format.
We need to use
for each section heading.
Inside each section, we need to use
for prose, and
- for any list.
Thus:
About the role
...
But the intro is multiple paragraphs; we can combine into one or multiple
. Probably multiple
okay.
We need to preserve all substantive information; we can split into paragraphs logically.
Let's craft About the role content:
First paragraph: "Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world."
Second paragraph: "DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior."
Third paragraph: "We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean."
Fourth paragraph: "This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders."
That covers the intro.
Now Responsibilities: "What You'll Do" bullet list.
We need to capture each bullet exactly.
From text:
"What You'll Do Set the research agenda Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and policies).Track emerging model architectures and specialized, domain-tuned model approaches, and translate what's relevant into DigitalOcean's product roadmap.Keep the agenda tightly coupled to product outcomes — every research bet should map to a measurable improvement in model selection, agent reliability, cost/latency, or task-success rate, not research for its own sake. Build and lead the team Grow the Applied Research team, hiring and mentoring research scientists and engineers.Establish the team's operating rhythm: how research questions get scoped, how experiments get run and evaluated, and how findings hand off to production teams.Represent Applied Research in cross-functional planning cycles alongside other engineering and product leaders, and make the case for headcount and investment on its own merits. Bridge research and product Partner directly with Inference Engine and the other platform and product engineering teams that own agent runtime and evaluation infrastructure to turn research into shipped capability.Turn research prototypes into production-ready capabilities in partnership with engineering — shipping research, not just publishing it.Communicate research trade-offs clearly to non-research stakeholders, including when a promising direction isn't ready for product investment yet. Represent DigitalOcean externally Build DigitalOcean's credibility in the applied agentic-AI research community through publications, talks, open-source contributions, or collaborations — where they serve product and hiring goals."
We need to split into bullet points. The original seems to have sections: Set the research agenda, Build and lead the team, Bridge research and product, Represent DigitalOcean externally. Each with sub-bullets.
We can create
- for each top-level bullet, and inside each li we can have
or just text. But instruction: use
- for any list of duties/requirements/benefits, and
for prose. So we can have
- containing prose (maybe
inside). Simpler: each li contains the text.
Thus:
Responsibilities
- Set the research agenda: Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and
- . For qualifications list also
- . For benefits also
- . For maybe skills? Not needed.
We need to preserve all substantive information; do not invent or summarize away detail. So we need to keep the exact wording? Probably we can paraphrase but must keep substance. Safer to keep the sentences as is, maybe break into paragraphs.
We need to remove boilerplate: EEO, equal-opportunity, scam-warning, application instructions, duplicated lines, cookie/navigation/legal. At end there is "DigitalOcean is an equal-opportunity employer..." and "Application Limit: You may apply to a maximum of 3 positions...". Those should be removed.
Also "JR: 2026-8125" is job ID; we should fold minor metadata into a short
or drop it. Probably drop it. Also "Compensation Range $249,600 - $312,000This is a hybrid role" appears as a line without space. We need to separate. Let's extract content. Original text: "Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world. DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior. We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean. This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders." Then "What You'll Do" with bullet points. Then "What You'll Add To DigitalOcean" with bullet points (required qualifications). Then "Preferred Qualifications" with bullet points. Then "Compensation Range $249,600 - $312,000This is a hybrid role JR: 2026-8125" Then "Why You’ll Like Working for DigitalOcean" with paragraphs. We need to produce sections: - About the role: maybe combine the introductory paragraphs up to "This is a hybrid role by design:"? But we also have hybrid role mention later. Could put About the role as the intro describing mission and role. - Responsibilities: the "What You'll Do" bullet list. - Requirements: the "What You'll Add To DigitalOcean" bullet list (required qualifications). Could also include "Preferred Qualifications" as separate section or under Requirements? The instruction says prefer few substantive sections over many thin ones. Could combine Required and Preferred into one "Qualifications" section with subheadings? But we only allowed headings list: About the role, Responsibilities, Requirements, Qualifications, Skills, Benefits, Pay, Schedule. So we could have Requirements for required qualifications, Qualifications for preferred qualifications, Skills maybe not needed. Or we could put both under Requirements and skip Qualifications. But we need to preserve all info; we can have both sections. - Benefits: the "Why You’ll Like Working for DigitalOcean" paragraphs. - Pay: compensation range. - Schedule: hybrid role. We need to ensure we don't create sections with filler. All have content. Now format. We need to use
for each section heading. Inside each section, we need to use
for prose, and
- for any list.
Thus:
About the role
...
But the intro is multiple paragraphs; we can combine into oneor multiple
. Probably multiple
okay. We need to preserve all substantive information; we can split into paragraphs logically. Let's craft About the role content: First paragraph: "Dive in and do the best work of your career at DigitalOcean. Journey alongside a strong community of top talent who are relentless in their drive to build the simplest scalable cloud. If you have a growth mindset, naturally like to think big and bold, and are energized by the fast-paced environment of a true industry disruptor, you’ll find your place here. We value winning together—while learning, having fun, and making a profound difference for the dreamers and builders in the world." Second paragraph: "DigitalOcean is building the Agentic Inference Cloud — anchored by Inference Engine, our model-serving platform — to help fast-growing AI-native companies start, scale, and optimize agentic workloads. Applied Research is the layer that generates the proprietary science that compounds across that platform: routing that learns instead of following static rules, memory that improves recall and personalization, observability that explains why agents succeed or get stuck, and reinforcement learning that closes the loop between production signals and model behavior." Third paragraph: "We're hiring a Director of Research to build and lead this function: set the research agenda, grow the team, and make sure the science this group produces ships into real product outcomes — better model selection and routing, more reliable agents, lower cost and latency, and higher task-success rates for the developers and enterprises running agentic workloads on DigitalOcean." Fourth paragraph: "This is a hybrid role by design: enough technical depth to personally evaluate and shape research direction across routing, memory, observability, and RL, and enough leadership range to build a team, prioritize against a roadmap, and defend research investment to product and executive stakeholders." That covers the intro. Now Responsibilities: "What You'll Do" bullet list. We need to capture each bullet exactly. From text: "What You'll Do Set the research agenda Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and policies).Track emerging model architectures and specialized, domain-tuned model approaches, and translate what's relevant into DigitalOcean's product roadmap.Keep the agenda tightly coupled to product outcomes — every research bet should map to a measurable improvement in model selection, agent reliability, cost/latency, or task-success rate, not research for its own sake. Build and lead the team Grow the Applied Research team, hiring and mentoring research scientists and engineers.Establish the team's operating rhythm: how research questions get scoped, how experiments get run and evaluated, and how findings hand off to production teams.Represent Applied Research in cross-functional planning cycles alongside other engineering and product leaders, and make the case for headcount and investment on its own merits. Bridge research and product Partner directly with Inference Engine and the other platform and product engineering teams that own agent runtime and evaluation infrastructure to turn research into shipped capability.Turn research prototypes into production-ready capabilities in partnership with engineering — shipping research, not just publishing it.Communicate research trade-offs clearly to non-research stakeholders, including when a promising direction isn't ready for product investment yet. Represent DigitalOcean externally Build DigitalOcean's credibility in the applied agentic-AI research community through publications, talks, open-source contributions, or collaborations — where they serve product and hiring goals." We need to split into bullet points. The original seems to have sections: Set the research agenda, Build and lead the team, Bridge research and product, Represent DigitalOcean externally. Each with sub-bullets. We can create
- for each top-level bullet, and inside each li we can have
or just text. But instruction: use
- for any list of duties/requirements/benefits, and
for prose. So we can have
- containing prose (maybe
inside). Simpler: each li contains the text. Thus:
Responsibilities
- Set the research agenda: Define and own the applied research roadmap across adaptive routing (evolving model selection from static rules into a system that learns from real usage, cost, and latency), memory (retrieval quality and durable recall for long-running agents), agent observability (understanding when agents make progress, get stuck, or make mistakes), and reinforcement learning / closed-loop learning (turning production feedback into better models and
- containing prose (maybe
- for any list of duties/requirements/benefits, and
- for each top-level bullet, and inside each li we can have
- for any list.
Thus:
- . For maybe skills? Not needed.
We need to preserve all substantive information; do not invent or summarize away detail. So we need to keep the exact wording? Probably we can paraphrase but must keep substance. Safer to keep the sentences as is, maybe break into paragraphs.
We need to remove boilerplate: EEO, equal-opportunity, scam-warning, application instructions, duplicated lines, cookie/navigation/legal. At end there is "DigitalOcean is an equal-opportunity employer..." and "Application Limit: You may apply to a maximum of 3 positions...". Those should be removed.
Also "JR: 2026-8125" is job ID; we should fold minor metadata into a short
- . For benefits also