Principal Engineer at Syfe (2021-02 – Present)
Built Syfe's brokerage backend from the ground up in a lean 5–10 engineer team — now one of the company's largest revenue lines. Owned architecture end-to-end: market-data (Spring WebFlux, Cassandra, Redis) and order/execution/settlement (Spring Boot, PostgreSQL, Kafka) microservices.
- Signature architecture bet — built mark8, the reactive (Spring WebFlux) market-data service with three-layer caching (in-memory, Redis, Cassandra) at <50ms latency — sustaining far higher traffic on fewer resources than a blocking design; exposes REST, WebSocket & SSE endpoints for ~10K DAU / 30–50K MAU.
- Scaled core APIs from ~500 → 3000+ peak RPS (5–7M calls/day per service) at P99 <300ms; lifted throughput ~40% and cut page/API latencies 50–80%+ via indexing, response-shape optimization, Protobuf, fewer internal hops, and resource tuning.
- Cut infrastructure cost ~50% — faster APIs shrank the pod footprint at peak, alongside auto-scaling and safe Spot-instance usage.
- Drove reliability as a discipline, not just response: defined SLOs/error budgets, designed for multi-region/DR and correct consistency under partial failure, and led failure-mode hardening — backed by 5–15 min MTTA/MTTR and structured RCAs.
- Owned software supply-chain security: Snyk-driven CVE remediation, dependency hygiene, and staying current on LTS versions to keep the system free of known vulnerabilities.
- Designed an isolated adapter layer for broker/bank/vendor integrations (e.g. HSBC, JPMorgan) over FIX, REST, WebSocket, SFTP, SDKs — idempotent, resilient flows holding correct state through vendor outages.
- Delivered price alerts, auto-invest (recurring orders) & auto-FX; built/automated per-order and net settlement with brokers, ensuring reconciliation and auditability of all cash & position movements.
- Led PostgreSQL optimization (partitioning, indexes, query-plan analysis) and Kafka tuning for high-throughput topics (consumer lag occasionally ~100k msgs/min); built a Kotlin multi-module Gradle architecture enforcing clear boundaries for clean microservice extraction.
- Introduced org-level tooling: shared Kafka configs, parallelized unit tests + coverage in GitHub Actions, k6-based load testing.
Senior Software Developer at JTG (2016-06 – 2021-01)
Architected and delivered backend services & REST APIs end-to-end (design → production) using Python (Django, DRF), leading teams of 3–8 engineers.
- Built high-performance APIs integrating Kafka, RabbitMQ, Celery beat, Redis, and AWS to meet strict latency and reliability requirements.
- Took full ownership with a strong bias for action: drove incidents to closure, fixed bugs at root cause, improved reliability through observability and safeguards.
Technical Leadership & Org Impact at Syfe (2021-02 – Present)
Member of the engineering leadership group; set technical architecture direction across all 5–6 product teams.
- Member of the engineering leadership group; set technical architecture direction across all 5–6 product teams — design decisions, build-vs-buy calls, and standards beyond my own services.
- Pioneered agentic engineering org-wide: built a backend dev-workflow plugin — adopted across all teams — that automates the full SDLC as a spec-first pipeline (specs as the source of truth): code generation via purpose-built skills, a multi-phase self-review (code quality, spec compliance, and security) enforced by both deterministic guardrails (Detekt, ArchUnit) and skill-based AI guardrails, automated test enhancement to >95% branch coverage, UAT deployment + validation, and PR creation with smart reviewer routing by team load. Introduced a consistent, formal dev process that cut development time and standardized PR quality across the org.
- Mentored engineers across SDE 1, 2, and 3; recognized go-to for distributed systems, performance, and design.
- Conducted 250+ interviews across backend and full-stack at multiple levels; helped shape and uphold the hiring bar.
- Established and strengthened RCA/postmortem culture to systematically reduce recurring incidents.
