Open to new-grad SDE / MLE roles

Backend systems and
ML infrastructure, built from scratch.

I build distributed systems and ML pipelines from the ground up — not with frameworks that hide the mechanism, but with hand-built storage engines, replication protocols, and retrieval algorithms I can defend line by line.

Explore the work Get in touch
244k
records/sec · mini-kafka
2000+
QPS · RAG retrieval
97%
recall@10 · 6M vectors
<$0.10
per review · DecisionDesk
01 Selected work
Go · distributed systems

mini-kafka

Live Source

A distributed, partitioned, replicated commit log — hand-built from the disk up. Segmented append-only log with sparse index and CRC, ISR replication, epoch-based leader election, idempotent producer, and group-commit batching. Zero external dependencies on the data path.

183–244k rec/s throughput P99 <18.4ms @ acks=all 71 tests · race-clean 0 external deps
GoTCP binary protocolsegmented logISR replicationleader electiongroup commitDockerGitHub Actions
Python · ML systems

RAG Engine

Live Source

A benchmark-proven retrieval-augmented generation system over a 6M+ vector Wikipedia corpus. Hand-profiled with py-spy — the flamegraph exposed a batching bottleneck, yielding a 3× QPS gain. Agentic multi-hop reasoning (iterative retrieve-reason-retrieve with self-reflection) lifts HotpotQA exact-match by +15 points over single-shot RAG.

2000+ QPS P99 <5ms ANN search 97% recall@10 93% faithfulness ~$0.004/query
PythonFAISS HNSWFastAPILangGraphClaude APIBM25 hybridBEIR / HotpotQAPrometheus
Multi-agent · LLM orchestration

DecisionDesk

Live Source

A real-time multi-agent architecture review system. Eight specialised LLM agents — planner, four parallel specialists (scalability, security, cost, maintainability), conflict detector, critic, synthesizer — run concurrently via LangGraph fan-out and stream verdicts over WebSocket. A human-in-the-loop revision loop re-runs the synthesizer without re-invoking the critic.

1.8× faster vs sequential 8 parallel agents <$0.10 per review 3.5× cheaper routing
PythonLangGraphFastAPIWebSocketOpenAI Responses APIPydanticPrometheusGrafana
02 What I work on

Systems depth

  • Distributed storage engines — segmented logs, WAL, CRC checksums, crash recovery
  • Replication protocols — ISR, high-watermark, epoch fencing, log truncation
  • Leader election — term-based, controller architecture, failover <5s
  • ANN indexing — HNSW, IVF-PQ, benchmarked at 6M vectors
  • Binary TCP protocols — framing, batching, backpressure
  • Multi-agent orchestration — LangGraph parallel fan-out

Production practice

  • Go — goroutines, channels, race-clean, golangci-lint, pprof
  • Python — FastAPI, Pydantic, mypy strict, pytest / hypothesis
  • Observability — Prometheus, Grafana, OpenTelemetry, SLOs
  • CI/CD — GitHub Actions, Docker multi-stage, Hetzner deploys
  • Testing — property-based, deterministic fault injection
  • ML serving — RAG, reranking, eval-gated pipelines, Claude API
03 Get in touch

Let's build something durable.

Targeting new-grad SDE / MLE roles. Always happy to talk systems.

Email GitHub LinkedIn