Realtime Captioning (Prod)
Low-latency ASR pipeline, streaming inference, and a React control panel for live monitoring. Deployed on K8s with autoscaling.
Python • RTMP • GPU • Prometheus
AI Full‑Stack • Machine Learning • Production
Kundan Yadav — AI Full Stack Developer
I build scalable AI systems, production-grade ML microservices and delightful front-ends. I bridge models and products: APIs, pipelines, inference infra, and user‑facing apps.
Open to: Remote/Contract
Location: India
KY
5+ yrs
I design model-first systems and ship them to production: inference endpoints, observability, CI/CD for ML, and responsive frontends.
Python • Node • FastAPI
React • Three.js • GSAP
K8s • Docker • Terraform
About
I combine deep learning knowledge with engineering practice. I build robust inference endpoints, monitoring, and intuitive frontends to expose model capabilities to users. I focus on reliability, latency, and delightful UX.
Model Deployment & CI/CD
Realtime Inference & Optimization
User‑centric Frontends
Skills & Tools
Backend & APIs
FastAPI, Node, DDD, REST/GraphQL, gRPC
Machine Learning
PyTorch, Transformers, Fine-tuning, Serving
Frontend & UX
React, Three.js, GSAP, Tailwind
Infra & DevOps
K8s, Docker, Terraform, CI/CD
Data & Pipelines
ETL, Feature Stores, Airflow
Testing & Monitoring
Prometheus, Grafana, SLOs
Selected Projects
Multimodal Search
Vector search service that fuses image & text embeddings, with a fast API and a polished front-end experience.
FAISS • FastAPI • React
Personal Agent SDK
SDK for building task-specific agents with model orchestration, retry policies and unified metrics.
Node • Docker • CI
3D Visualizer
Interactive Three.js visualizations for model introspection and embeddings projection with animated transitions.
Three.js • React
Contact
Let's build together
Available for remote roles and freelance. Typical engagements: prototype → production.
Email: kundan@example.com
LinkedIn: linkedin.com/in/kundan-yadav
GitHub: github.com/kundany