Aditya Katkar

I’m a Software Development Engineer with 2+ years owning end-to-end design, deployment, and operation of production LLM inference pipelines and distributed systems for global enterprise customers. I’m driven by Customer Obsession, automating workflows that previously required human cognition across vendor evaluation, compliance monitoring, and forensic investigation.

I’ve owned a 12-microservice Kubernetes platform processing 10,000+ daily AI jobs with 99.7% uptime. The work delivered measurable outcomes: 69% P95 latency reduction, 22% cloud cost savings, 98% Firestore read elimination, and collapsing 4-week procurement cycles to same-day processing.

My engineering approach blends deep system design with pragmatic reliability. I’ve productionized multiple LLM backends (Qwen2-VL, vLLM, Gemini 2.5 Flash) with per-task routing, fallback logic, and safe production access controls, while tackling GPU overload and retry storms using health-check circuit breakers and priority-based orchestration.

I also enjoy building platforms that make complex decisions auditable and usable—turning slow, error-prone processes into dependable pipelines. From Django ORM performance tuning to Redis caching and graph intelligence with Neo4j, I focus on reliability, measurable customer impact, and mentoring teammates.