Birat Gautam

Experience & Education

1. N
   NepaWorksAI/ML Engineer

   Aug 2026 – Present#e6d2b74

   • Architected a distributed lead-classification platform across three services — async FastAPI control plane on PostgreSQL, containerized Python worker fleet, and a React operator console — processing 4,000+ businesses end-to-end at 99.95% completion.
   • Deployed a 12B open-weight vision-language model on-premise via llama.cpp on RTX 5090 hardware, classifying businesses from scraped imagery and reviews at ~650/hour with zero per-token API cost.
   • Designed a leader-elected scheduler using transaction-scoped PostgreSQL advisory locks and lease-based work claiming, enabling safe horizontal scaling of the API under pgbouncer transaction pooling.
   • Built a stateless worker fleet with heartbeat-driven health aggregation, server-side work gating, and exponential backoff, sustaining multi-hour unattended runs with zero failed rows and full SIGKILL crash recovery.
   • Implemented Server-Sent Events streaming over an append-only PostgreSQL event log with cursor-based resume, delivering gap-free realtime pipeline updates across service redeploys.
   • Engineered a concurrent Playwright scraping subsystem that harvested 27K+ images and 31K+ reviews, applying fingerprint hardening and calibrated concurrency to run block-free at production volume.
   • Optimized the pipeline through profiling: 24x faster export queries (448ms to 18ms), 90% reduction in image bandwidth, and elimination of memory and process leaks that were terminating multi-hour runs.
   • Reduced manual qualification of 3,000+ prospects to a sub-1% human review queue by persisting model reasoning and cited image evidence for every classification.
2. S
   Synapse TechnologiesJr. AI/ML Engineer

   Nov 2025 – Jul 2026#a3f91c2

   • Engineered a multilingual NLP pipeline for English, Nepali, and Romanized Nepali queries using transformer-based intent classification and script-aware NER; reduced inference latency from ~10s to ~30ms via ONNX optimization and INT8 quantization.
   • Developed and productionized the KanoonBox RAG pipeline using ChromaDB vector search and NepSearch-based legal grounding; optimized intent-routing architecture to reduce reasoning latency from ~5s to ~25ms while maintaining response quality.
   • Deployed Nepali document OCR using PaddleOCR-VL-0.9B via llama.cpp within 2 GB VRAM (7s/page); integrated Qwen2.5-VL-3B-Instruct-GGUF into Synapse Intelligence for enterprise document understanding and knowledge delivery.
   • Built Nepal's first structured Nepali ASR corpus from YouTube audio using VAD, NFC normalization, and Gemini validation; LoRA fine-tuned Whisper-large-v3-turbo on resource-limited hardware (WER: 33, WandB monitored).
   • Benchmarked lightweight Gemma 4 variants on CPU infrastructure using Ollama, llama.cpp, and GGUF quantization formats for on-device LLM deployment feasibility.
3. B
   Birmingham City UniversityBSc (Hons) CS with AI

   2023 – Present#7b2d441

   • Currently Third Semester Student.
   • Spearheaded a team of five to build a face-recognition attendance web app.
   • Completed First Year with First Class Honours.
4. D
   Deerwalk Compware Ltd.Backend Developer Intern

   Jun 2023 – Aug 2023#4e8c910

   • Designed and developed an Attendance Management System using Django.
   • Implemented Excel report generation for detailed attendance records.
   • Built an Admin Panel with full CRUD operations.
5. U
   Uniglobe SS/CollegeHigh School Graduate

   2020 – 2022#1f2a398

   • Graduated with a 3.78 / 4 GPA.
   • Awarded Student of the Year — UGSS Achievers Award 2022.