I build backend services, real-time applications, and systems that connect research with practical engineering. My work spans Python and Go APIs, durable workflows with Temporal, Kubernetes infrastructure, and local AI.
B.Tech in Computer Science & Engineering, IIIT Guwahati (2022-2026) · CGPA: 8.48/10
Back End Developer Intern · GRIG Technologies
June 2024 - August 2025 · Remote
- Built alert.ventory with FastAPI and Temporal, combining priority queues, retries, and human-in-the-loop approval for enterprise notifications.
- Developed request.ventory, a multi-tenant inventory API with Firestore and Supabase, including GST and geographic data validation.
- Implemented RBAC, Zoho OAuth2, and JWT session management for internal services.
- Automated Kubernetes setup with Helm, NGINX Ingress, and cert-manager, reducing environment setup time by 90%.
- Built CI/CD pipelines with GitHub Actions and Docker, including deployments to Google Cloud Run.
Go · Gin · PostgreSQL · WebSockets · Vue 3
A multiplayer geography game with synchronized lobbies, server-side Haversine scoring, and a Go worker pool that validated 81,000+ Street View locations. The frontend uses Pinia to coordinate game state and map interactions.
FastAPI · React · Firebase · WebRTC · Docker
A real-time chess variant with custom portal movement, synchronized timers, and PeerJS voice chat. The backend provides ELO analytics, Firebase/JWT authentication, asynchronous Firestore access, and transactional friendship operations, with automated deployment to Google Cloud Run.
Python · Faster-Whisper · SentenceTransformers · Ollama · Piper
A local-first voice assistant with embedding-based intent matching, local LLM fallback, and custom CRF entity extraction achieving a weighted NER F1 of 0.9977. Inference workers and queued speech synthesis run separately from the UI.
PyTorch · Double DQN · MADDPG · LSTM · SUMO
A hierarchical reinforcement learning framework for UAV fleet sizing and positioning across 8 simulated cities. Predictive caching trained on Azure workload traces improved cache hit ratio from 18% to 78% and reduced average task latency by 53% in simulation.
- Fog Computing Testbed: A four-tier Docker testbed for IoT module placement, with Linux network emulation, Redis streaming, and Prometheus/Grafana observability.
- Pilgrim AI: An earlier Flutter and Flask conversational AI project using Vertex AI, session-based chat context, and Google App Engine.
| Area | Technologies |
|---|---|
| Languages | Python, C++, SQL, Go, JavaScript |
| Backend | FastAPI, Flask, Gin, Temporal, SQLAlchemy |
| Frontend | React, Vue 3 |
| AI & ML | PyTorch, NumPy, SentenceTransformers, Faster-Whisper, Ollama |
| Cloud & Data | Google Cloud, AWS, PostgreSQL, Redis, Supabase, Firebase, MongoDB |
| Infrastructure | Docker, Kubernetes, Helm, GitHub Actions, Prometheus, Grafana, Linux, Git |
Explore the architecture, demos, and research reports at rudragarg.dev.