Backend Engineer specializing in scalable distributed architectures, asynchronous data pipelines, and production AI/ML integrations. What I work on: • Core Backend & Systems: Designing resilient microservices using Python (FastAPI, SQLAlchemy, Celery, Redis). • Infrastructure & Cloud: Containerized orchestration across Kubernetes, Docker, Helm, and KEDA, enforcing multi tier ingress routing and domain isolation with Traefik and ArgoCD. • Algorithms & Optimization: Building constraint programming solvers and reconciliation engines (Google OR-Tools, NetworkX) for complex financial and operational workflows. • Applied AI & ML: Architecting end-to-end multimodal document parsing and LLM inference pipelines with structured output schemas. Published researcher in NLP/ML (Natural Language Processing Journal, 2025). Always open to discussing distributed systems design, backend throughput optimization, and production grade engineering opportunities.
AI & Software Engineer
Cappitall Want
- • Combinatorial Optimization Engine: Engineered an automated financial matching engine using Google OR Tools (CP-SAT), evaluating millions of constraint combinations to reduce manual audit reconciliation from 14 days to under 10 minutes with 100% mathematical precision.
- • Autonomous Email Ingestion & Invoice Pipeline: Architected an extensible, provider agnostic email automation engine utilizing abstract base interfaces across Gmail and Outlook. Orchestrated an LLM driven classification agent to evaluate body context and attachments for downstream extraction.
- • Distributed Task Architecture & Worker Scaling: Deployed a scalable Celery execution backbone on Kubernetes with persistent event loops and queue specific resource quotas, isolating heavy compute workloads (GST/TDS extraction) from core API services.
- • Asynchronous System Architecture & Security: Architected an enterprise grade FastAPI microservices framework, establishing 3-tier Layer-7 route isolation to restrict internal cluster endpoints at the Ingress controller while eliminating redundant pod to pod auth overhead.
- • Multimodal Parsing & Compute Optimization: Engineered an adaptable multimodal document extraction engine. Migrated single threaded bottlenecks to Python multiprocessing pools, slashing end to end processing latency by 90%.
- • Stateful Agent Orchestration Platform: Built a centralized factory framework for scalable AI agent execution, abstracting state management, system instructions, and dynamic tool calling for low latency LLM tasks.
Attendee · Oct 24, 2026