Headline

Building production AI systems. Multi-agent backends, voice agents, and open-source tools with 20K+ downloads. AI Engineer @Iamreadyai | YC W26

About

I used to think the hard part of AI was getting the model to converge. Then I shipped my first production system and realized the hard part is everything else. The 4 AM alert when Redis cache expires. The voice agent that has to sound human enough that people do not hang up. The STT model that chokes on background noise. The workflow builder that non-technical clients need to use without breaking the internet. Now I build systems that survive all of that. I am currently leading Applied AI at Sasefied, where I shipped IamreadyAI end-to-end. Five enterprise clients later, I had a CTO and Co-founder offer and a platform handling 10K-plus daily requests. I also architected the Sasefied workflow builder, a drag-and-drop canvas for composing AI agent systems that deploys via a Python SDK or a single-line script embed. When latency became a bottleneck, I cut response time by 15 percent and scaled the backend with RabbitMQ async workers. Before that, I was at Samora AI, a Y Combinator W26 company, building voice agents for e-commerce. Think abandoned cart recovery through actual mobile calls and WhatsApp messages. I instrumented production pipelines with LangSmith, benchmarked speech-to-text models on live call audio, and learned that production AI is 20 percent model and 80 percent infrastructure. Outside of work, I publish open-source tools that developers actually use. 20K plus PyPI downloads and counting. MCP-Arena, a modular framework for production MCP servers, is the one I am most excited about right now. Stack I work with: AI and ML: PyTorch, TensorFlow, Scikit-learn, LangGraph, LangChain, Hugging Face, RAG, AI Agents, MCP LLM Fine-Tuning: QLoRA, LoRA, PEFT, Transformers Backend: Python, FastAPI, PostgreSQL, MongoDB, Redis, RabbitMQ Frontend: Next.js, TypeScript, Vite DevOps and Infra: Docker, AWS, GitHub Actions, CI/CD, Vercel, Render When I am not chasing p99 latency or debugging agent orchestration, I am usually on LeetCode. Knight tier, rating 1934, 400+ problems solved. Or building something I probably should not be building at 2 AM. I am currently open to AI engineer, Applied AI, and full stack AI roles, especially at startups where production-ready is the starting point, not a future milestone. B.Tech CSE at NIT Nagaland. CGPA: 9.80

Location

Gorakhpur

AI stack

anthropic apis

antigravity

claude code

claude

codex

cursor

Work experience

Founding AI Engineer at Iamreadyai, Jul 2026 - Present · 2 mos, Remote, • Built an automated client finder using a customized LangChain agent harness with custom skills, surfacing 10–15 qualified leads per chat while reducing agent error rate from ~40% to under 10% via LangSmith observability and iterative refinement. • Cut enterprise client onboarding from 4–5 days to under 1 day by building dedicated account provisioning endpoints; personally onboarded 8+ clients with same-day deployment. • Architected a RabbitMQ async pipeline with 5 concurrent worker threads to offload resume/job processing (parsing, embedding generation, vector indexing) from the main thread, eliminating request-blocking and improving backend throughput.

Founder at DynamisOS, Jul 2026 - Present · 2 mos, Remote, • Founded DynamisOS, a startup focused on developing AI that personalizes software experiences. • Enabled users to become their own Forward Deployed Engineers (FDEs), tailoring software to individual needs.

Founding Engineer at Agent Run, Apr 2026 - Jul 2026 · 4 mos, Remote, • Implemented a Self-Healing AI Agent System using LangChain, LangGraph & LangSmith—where agents automatically reflect on failures, correct their approach, and retry without human intervention, and every decision is fully traced and observable. • Published a technical blog on AetherMind's autonomous agent architecture, documenting the design behind self-healing workflows, hash-pointer memory routing, and 16-shard context management for production-grade agent systems.

Software Engineer (Applied AI) at Sasefied, Jul 2025 - Aug 2026 · 1 yr 2 mos, • Shipped IamreadyAI, an AI-powered hiring platform with embedded careers sites and AI interview prep; onboarded 5+ enterprise clients and earned CTO & co-founder offer. The platform serves 10K+ daily requests with JWT authentication, Redis caching, rate limiting, and 4-level RBAC. • Architected Sasefied workflow builder—a visual drag-and-drop canvas for composing AI systems via configurable nodes. Delivered dual integration via Python SDK or single-line script embed. Cut response time by 15%. • Designed a talent acquisition agent using Google Search, LinkedIn, and GitHub profile matching. Achieved ~96% success rate and lowered p99 latency. • Scaled backend with RabbitMQ async workers, eliminating request-blocking on long-running tasks.

Software Development Intern at Sasefied, Apr 2025 - Jun 2025 · 3 mos, • Built AI-powered resume parsing and enhancement system using LLMs for grammar correction, keyword optimization, and semantic alignment with job descriptions. • Engineered resume-job matching using embeddings and a real-time chatbot service for candidate/recruiter queries. • Designed criterion-based interview feedback system (communication, technical, problem-solving) with Plotly visualizations and Redis-backed storage. • Developed multi-tenant Node.js backend with custom RBAC (4 hierarchical roles), JWT auth, and strict client-level PostgreSQL isolation. • Built Talent Acquisition Manager (TAM) Agent extracting public profiles from Google, LinkedIn, and GitHub for relevant candidate suggestions.

Education

Bachelor of Technology - BTech, Computer Science, National Institute of Technology Nagaland

Public links

github: https://github.com/SatyamSingh8306

leetcode: https://leetcode.com/aicinema69

youtube.com: https://youtube.com/@codebysatyam8306

codebysatyam.vercel.app: https://codebysatyam.vercel.app/