Principal Engineer| Hybrid: Chicago, IL

Job Responsibilities

• Architecture & Design: Play a key role in core architectural choices and roadmap for our distributed database kernel, including query execution engines, custom/hybrid columnar storage layers, distributed transaction managers, and cluster coordination protocols.
• Low-Level Development: Write, optimize, and maintain highly performant, production-ready backend code utilizing modern C++ (C++20/23).
• Parallel Execution Planning: Architect MPP execution engines that distribute, schedule, and execute queries across several dozens to hundreds of nodes using MPI and hybrid OpenMP models.
• Hardware Optimization: Profile and eliminate execution bottlenecks by implementing explicit SIMD vectorization (AVX-512, ARM Neon, SVE) and ensuring absolute cache locality (L1/L2/L3).
• Distributed Memory & Networks: Design ultra-low latency cluster communication subsystems utilizing RDMA, RoCE, or kernel-bypass tech (DPDK).
• In-Database ML Framework Design: Architect and implement a native, highly performant, zero-copy ML execution framework directly inside the MPP engine to eliminate data serialization and transport bottlenecks between the database and external AI pipelines.
• Functional & High-Performance Algorithms: Develop mathematically rigorous, cache-oblivious, and highly vectorized algorithms for real-time statistical computations, matrix operations/manipulations on clustered data.
• Vector & Embedding Infrastructure: Design and optimize or embed/reference distributed high-dimensional vector storage subsystems, custom SIMD-accelerated indexing structures (e.g., HNSW, IVF-PQ), and parallelized approximate nearest neighbor (ANN) search algorithms.

Required Technical Qualifications

• Minimum - 4 years Computer Science engineering degree from a reputed institute. Masters in HPC (high performance computing) and Bigdata engineering areas preferred for Senior staff engineer and above roles.
• Modern C++ Expertise: 12+ years of production experience writing hands-on, low-level systems in C++ (C++17/20/23 preferred). Mastery of template metaprogramming, concepts, coroutines, and custom memory allocators is key.
• HPC Parallelism: Very good to expert-level knowledge of cluster-scale distributed programming using MPI (Message Passing Interface) paired with shared-memory thread-level programming via OpenMP and MapReduce frameworks.
• Hardware-Aware Engineering: Documented experience with microarchitectural optimization, loop vectorization, pointer-aliasing constraints, and manual SIMD intrinsics programming.
• Concurrency Mechanics: Deep understanding of OS internals, memory barriers, lock-free data structures, and atomics (std::atomic).
• Database Internals: Practical experience designing core database subsystems (e.g., vectorized query operators, LSM-trees, B-trees, custom buffer pools, columnar serialization frameworks and query plan optimizers).
• Strong debugging, performance tuning, and problem-solving skills.
• Proficiency in one or more leading memory and compute monitoring/performance tools (such as Valgrind, Purify, Intel/AMD/ARM diagnostics etc.).

Bonus Qualifications

• Track record contributing to open-source or work experience on proprietary distributed engines (e.g., ClickHouse, DuckDB, Doris, CockroachDB, RocksDB, ScyllaDB, or Velox).
• Experience with cloud-native distributed, elastic environments, object storage tiering (S3/GCS), or zero-copy networking.
• Experience integrating heterogeneous acceleration (NVIDIA CUDA / ROCm) into big data workloads and hybrid CPU/GPU setups.
• Vector & Graph Algorithmic experience: Experience building or modifying high-performance vector index/search engines, graph-based indices, or localized quantization frameworks.
• Functional domain awareness – CPG, Retail, Pharma verticals