Capabilities
Engineering range for products that cross backend, AI, realtime and infrastructure.
I can own backend traffic, build the React or Next.js surface, put LLM agents into production, run GPU inference myself, debug the bottleneck, and keep the security risks visible.
01
Backend systems
Python, Go, Django, Django REST Framework, FastAPI, Celery, TaskIQ, Django Channels, WebSockets, gRPC, Pydantic, event driven architecture, and 10K requests per second workloads.
02
AI and LLM systems
LangChain, LangGraph, CrewAI, AutoGen, Claude API, OpenAI SDK, tool calling, multi agent orchestration, RAG over pgvector and Pinecone, embeddings, evals, structured outputs, and SLM fine tuning with LoRA and PEFT.
03
ML and generative
PyTorch, scikit learn, Hugging Face Transformers and Diffusers, CNNs, OCR. FLUX, LTX 2.3 and Qwen self hosted on RunPod A100 and H100 GPUs with ComfyUI and vLLM, tuned for cost per inference.
04
Frontend product surfaces
React, Next.js with App Router, SSR and ISR, TypeScript, Tailwind CSS, Zustand, Redux, React Query, admin dashboards and real time operator consoles scoped straight from product requirements.
05
Data and performance
PostgreSQL, Redis, MySQL, pgvector, PgBouncer, query plans, composite indexing, N+1 elimination, double entry ledgers, and a Rust task queue when Celery is the ceiling.
06
Cloud, DevOps and security
AWS EC2, S3, RDS, Lambda and CloudFront, RunPod, Docker, Caddy, Nginx, GitHub Actions, Azure DevOps, Linux administration, custom autoscaling, observability, Playwright, pytest, load testing, protocol reverse engineering and Django RCE analysis.
Engagement types
Ways I can plug into a team.
Job, contract or white label. The engineering is the same, the paperwork and the visibility change.
Full time or embedded
Join the team remotely from UTC+5, own a backend or an AI surface end to end, take the on call, and mentor through review.
Build from zero
Turn an idea into a working product with backend, frontend, database, deployment and operational structure, on fixed milestones with weekly demos.
White label delivery for agencies
I build under your brand, in your repo, to your standards. NDA by default, your client never sees me unless you want them to, and your work never appears on this site.
Recover or accelerate a system
Reverse engineer provider protocols, replace brittle vendor paths, find the bottleneck, and stabilise production without turning everything into a rewrite.
Add AI to an existing workflow
RAG, agents with eval harnesses, OCR, vector search, vision review, fine tuned small models, and self hosted generation when the vendor bill stops making sense.
Not sure which one you need?
Describe the problem, not the role. I will tell you which shape fits and what I would do first.