Quantitative Developer Internship (Winter 2026)

📋 Description

• • Own the full lifecycle of ultra-low-latency trading components—from white-board sketches to production deployment—using modern C++17/20 in a UNIX environment. You will profile, benchmark, and shave microseconds off critical paths that directly affect P&L and shape the firm’s competitive edge.
• • Design and extend our real-time market-data ingestion layer: parse exchange feeds (ITCH, OUCH, MDP, FIX/FAST), normalize messages, and fan them out through ZeroMQ/Kafka-style buses to downstream strategies with sub-millisecond latency budgets. Your code will be the first line of defense against stale or malformed ticks.
• • Build deterministic, event-driven simulation engines that let researchers replay billions of ticks and test new alphas without touching live capital. You’ll craft replay tooling, synthetic order books, and Monte-Carlo stress tests that run 24×7 on our GPU cluster.
• • Implement execution algorithms that split large parent orders into child slices while reacting to micro-structure signals—queue position, imbalance, spread, volatility. You’ll measure slippage, adverse selection, and fill probability in real time and iterate on the fly.
• • Instrument every service with fine-grained logging, metrics, and tracing so we can diagnose issues at 3 a.m. without waking the entire team. Expect to work with GDB, perf, flame graphs, Prometheus, Grafana, and custom eBPF probes.
• • Automate deployment pipelines that push new builds to AWS EC2 bare-metal instances and Docker containers with zero-downtime blue-green swaps. You’ll script the glue in Python or Bash, guard everything with unit/integration tests, and own the rollback strategy.
• • Collaborate shoulder-to-shoulder with PhD researchers to translate mathematical models into rock-solid C++ kernels. When a researcher says “factor X predicts 2 bps,” you make it run in 200 nanoseconds and prove it under live fire.
• • Triage and fix production incidents—whether it’s a memory leak, a kernel NIC driver quirk, or an exchange protocol change—under the pressure of live markets. Post-mortems are blameless but fast; root-cause fixes are permanent.
• • Contribute to our in-house libraries: lock-free queues, custom allocators, nanosecond clocks, SIMD intrinsics, and TCP bypass techniques. Your commits will be reviewed by engineers who previously built Tier-1 HFT stacks.
• • Present weekly lightning talks on performance tricks, new C++ features, or war stories from the trading floor. Teaching is the fastest path to mastery, and interns often become go-to experts by the end of the program.
• • Leave with a portfolio you can’t get in a classroom: order books you hardened, sim engines you accelerated, and strategies you helped launch with real capital. Your GitHub will sparkle with production-grade code that moved markets.