About Me

What I’m Doing now ?

• Software Engineer at Fynd, building Optimus: an autonomous commerce layer where AI agents plan, act and learn across a merchant’s whole business, with the merchant setting goals and approving the decisions that matter.

• Founder of Trueyy (trueyy.com), which detects AI-assisted cheating in live remote interviews and gives hiring teams a live integrity score on Zoom, Google Meet and Microsoft Teams.

• I am contributer of authorizer.dev , a database independent open-source authentication and authorization solution.

• Also contributer of ttpro.in which is textile suppliers management tool.

Experiences

• Fynd | Software Engineer, Agentic Commerce (Optimus) · Feb 2026 to Present
• Building the agent layer that lets AI operate an e-commerce business end to end. The merchant sets goals and approves the decisions that matter, and the agents handle the rest.
• Designed a multi-agent system in which a coordinator and domain specialists carry out real operations across catalog, orders, inventory, pricing, logistics, marketing and customer engagement, through one unified tool surface of 290+ operations.
• Solved the too-many-tools problem with a search, inspect and execute layer: the agent holds a small working set and pulls the rest on demand. External MCP clients now use the same pattern.
• Built a background planning layer that reads the merchant’s calendar and sales signals, works backwards from upcoming events, and prepares campaigns, stock and pricing days ahead. Only the decisions that need a human reach the merchant.
• Made safety a property of the architecture: human approval gates, signed write confirmations, deny-by-default auth and per-tenant scoping live at the tool and API layer, so the system stays safe regardless of what the model does.
• Built two interchangeable agent runtimes (LangGraph with Deep Agents, and the Claude Agent SDK) behind one streaming, tool, skill and tracing layer, so changing model vendor is a configuration change.
• Shipped an MCP server that exposes the full commerce platform to Claude, ChatGPT, Cursor and custom agents.
• Agents learn per-organisation skills from past runs and open knowledge files on demand. A self-healing loop diagnoses failures and opens draft fix PRs.
• Stack: TypeScript, Node.js, React, GraphQL Federation, LangGraph, LangChain, Claude Agent SDK, MCP, OpenAI and Anthropic APIs, PostgreSQL, pgvector, pg-boss, GCP, Kubernetes, Langfuse, OpenTelemetry

• Trueyy (trueyy.com) | Founder · 2026 to Present
• Founded Trueyy, an interview integrity platform that detects AI-assisted cheating in live remote interviews on Zoom, Google Meet and Microsoft Teams. Candidates join by a link and consent up front, with nothing to install. Interviewers get an integrity score that recalculates every 30 seconds, a timestamped timeline they can scrub back to, and a one-click summary that attaches to the ATS scorecard (Greenhouse, Lever, Workday).
• Studied how ChatGPT, Claude, Gemini, Cluely and InterviewCoder shape a live answer and built AI tool fingerprinting that reads their structural output signatures and prompt-and-paste timing. A single odd moment stays in context, and only the pattern gets logged.
• Six detection layers: AI tool fingerprinting, app and window focus, paste velocity, off-screen device signals (audio artifacts, ambient light, timing gaps), answer structure analysis, and a planned reading-gaze layer. The layers are weighed together, so a nervous pause passes and a sustained lookup gets flagged, with every flag tied to a timestamp.
• Consent is enforced in code: capture is gated on candidate consent and stops on revoke, session end or a lapsed heartbeat. JWT auth with refresh-token rotation, tenant-scoped role-based access, TLS in transit and AES-256 at rest, an immutable audit log, configurable retention with automated deletion and candidate erasure. The meeting video never leaves Zoom, Meet or Teams. GDPR aligned, SOC 2 in progress.
• Own the whole product: customer discovery, positioning against exam proctoring tools (Proctorio) and AI interviewers (Fabric), pricing tiers and an SDK for staffing agencies, docs, a library of buyer and detection guides, the website and go-to-market, alongside the engineering.

• Praalak Tech Solutions | Software Engineer (previously). Fashion supply chain traceability platform with React, Next.js, Node.js, Golang, Java Spring Boot, GraphQL and AWS.

• Contentment Foundation | Software Engineer (previously). Digital solutions for global wellbeing education, built with React.js, Next.js, Node.js, Golang and Java Spring Boot.

• Software Engineer for the Lakahn Samani technoloy

• Associate Software Engineer for the Softsages technology

• Associate Software Developer for the Car Stream (Freelance)

Education

• Dharmsinh Desai University, B.Tech in Computer Engineering (Aug 2019 to Mar 2023)

• CPI: 8.32

Skills

• Claude Agent SDK

• Model Context Protocol (MCP) server and client design

• Meta-tool layers (search / inspect / execute over large tool catalogs)

• Human-in-the-loop agent workflows (LangGraph interrupts, approval gates)

• Agent safety: signed write confirmations, deny-by-default auth, per-tenant scoping

• Langfuse, OpenTelemetry, LangSmith (agent tracing and cost attribution)

• pgvector, Prisma, pg-boss

• Kubernetes, Docker, GCP Cloud Run, BigQuery

• Apollo Server / GraphQL Federation

• Playwright, Vitest

• Golang

• NodeJs

• Java Spring Boot

• Javascript

• ReactJs

• NextJs

• GraphQL

• SQL

• NoSQL

• AWS

• TypeScript

• PostgreSQL

• MongoDB

• Firebase

• LangGraph

• Deep Agents

• LangChain

• MCP (Model Context Protocol)

• Socket.io / WebSockets

• Docker

• GCP / Cloud Run

• Python

• Multi-Agent Systems

• Agent Orchestration

• Task Decomposition

• LLM Engineering

• Prompt Engineering

• Prompt Chaining

• Retrieval-Augmented Generation (RAG)

• Tool Calling / Function Calling

• Context Management

• Memory Systems (AI Agents)

• Vector Databases

• Embeddings

• Agent Evaluation & Observability

• Guardrails & Safety

• Distributed Systems

• Microservices Architecture

• Event-Driven Architecture

• Async Programming

• API Development (REST/GraphQL)

• System Design

• Queue Systems (Kafka / Redis / RabbitMQ)

• Cloud Computing (AWS / GCP / Azure)

• DSA ( 500+ problem solved in Leetcode )

What I enjoy

• Learning new technologies.

• Reading Books

• Playing outdoor games to keep my self fit.

• Listening music