Ajay Adkar
Low-Latency Systems & C++ Engineer
Low-Latency Systems & C++ Engineer with 10+ years of experience designing and optimizing deterministic execution engines, exchange connectivity gateways, and high-throughput market data pipelines for capital markets. Expert in C++ (up to C++23), Linux systems programming, lock-free concurrency, and memory management. Specialized in eliminating microarchitectural bottlenecks and latency jitter using kernel bypass (AF_XDP) and real-time operating systems (PREEMPT_RT). Proven track record of bridging low-level systems engineering with quantitative infrastructure to build highly reliable, sub-microsecond trading systems.
AF_XDP and XDP/eBPF pipelines on AWS Graviton3 — packets redirected to userspace before the kernel networking stack.
Order execution engines, exchange connectivity, OMS, orderbook/tradebook systems, and backtesting platforms at scale.
AWS Bedrock + LLM orchestration for quantitative research workflows, semantic market pattern parsing, and financial analytics tooling.
Live Infrastructure Demo
AWS Graviton3 (c7g.xlarge) · AF_XDP kernel-bypass · ENA · UDP:9001 · 1 s telemetry
Career Highlights
- ✦Architected asynchronous, event-driven trading engines and exchange connectivity platforms in Modern C++, ensuring minimal critical-path overhead.
- ✦Designed and benchmarked a high-throughput market data ingestion prototype utilizing AF_XDP kernel bypass to achieve zero-copy packet transfer.
- ✦Eliminated latency spikes and processing jitter by deploying systems on real-time Linux kernels with strict core affinity and isolated CPUs.
- ✦Founded and engineered a high-throughput quantitative backtesting engine designed to process massive historical datasets and tick-by-tick simulations with minimal computational overhead.
- ✦Re-engineered core financial backend workflows using lock-free ring buffers, custom memory pools, and multi-threaded event pipelines.
- ✦Collaborated directly with algorithmic traders and quantitative researchers to transition complex financial analytics and LLMs into low-overhead workflows.
Technical Skills
Languages & Standards
Low-Latency & Systems
Trading & Market Infrastructure
Backend & Data
Platforms & Tools
Professional Experience
Low Latency Trading Engine
- ›Architected asynchronous, event-driven backend APIs using C++17/20, Boost, and the Drogon framework to orchestrate user order placements, cancellations, and real-time state tracking.
- ›Developed low-overhead socket layers to manage direct exchange communication protocols independently of the client-facing web framework.
- ›Optimized backend API throughput and message handling by leveraging non-blocking asynchronous I/O, optimized thread pools, and high-frequency WebSocket streams.
- ›Shifted runtime overhead to compile-time by replacing heavily dynamic STL allocations with fixed-size structures, custom memory management, and std::string_view to mitigate latency jitter.
- ›Developed and unified high-throughput infrastructure components across transaction pipelines and multi-asset trading applications.
Tradebook & Orderbook Reporting Platform
- ›Engineered highly efficient C++ APIs for orderbook and tradebook reporting systems, optimizing data streaming and parsing routines to process large historical transaction dumps with minimal memory footprint.
- ›Scaled multi-threaded query execution layers handling historical trade audits by anchoring pipelines onto decoupled, event-driven Kafka message streams.
- ›Designed highly efficient binary data-serialization and streaming workflows to process massive financial logs without locking the main execution thread.
Financial Analytics & Research Tooling
- ›Built high-throughput Python/FastAPI microservices tailored for real-time stock analytics, growth calculations, and quantitative research workloads.
- ›Created scalable analytical layers utilizing vectorized NumPy and Pandas processing pipelines to execute complex, high-throughput historical portfolio valuation and risk checks.
- ›Integrated AWS Bedrock LLM capabilities into trading research tools, automating semantic market pattern parsing and increasing researcher extraction throughput.
Quantitative Backtesting Engine
- ›Founded and developed a high-throughput quantitative backtesting engine from scratch using C++ to execute tick-by-tick simulations over historical market data.
- ›Implemented multi-threaded simulation loops pinned to isolated CPU cores to run concurrent historical strategy parameter valuations without thread-switching overhead.
- ›Engineered high-speed data loaders utilizing asynchronous I/O to ingest historical market datasets into memory with minimal CPU overhead.
- ›Created highly parallelized data parsers capable of computing portfolio drawdowns, transaction costs, and statistical strategy metrics simultaneously.
Mutual Fund Trading Platform
- ›Developed high-availability Python services to process mutual fund order-routing workflows, managing complex state machines across transactional boundaries.
- ›Optimized PostgreSQL database schemas, indexing strategies, and connection pooling to minimize API blocking under high-concurrency order placement loads.
- ›Tuned Linux-based runtime environments to ensure deterministic processing speeds and 99.99% uptime during peak Indian market hours.
Real-Time Market Data & Trading Application
- ›Engineered a real-time market data feed handler using WebSocket connections to ingest, parse, and broadcast live exchange tick updates directly to front-end systems.
- ›Integrated low-overhead connectivity layers linking Order Management Systems (OMS) with funds management validation modules.
- ›Reduced data-handling latency by streamlining payload serialization, decreasing application response times during sudden market volatility spikes.
Core Platform Development
- ›Contributed to core platform features, bug resolution, and performance improvements across trading application workflows.
Android KYC & Client Onboarding
- ›Assisted in maintaining and enhancing an Android-based client onboarding and KYC application, taking ownership of smaller modules and independently implementing new functionality.