Full-stack · AI · Product
I build AI-powered products and the full-stack systems behind them.
Work includes Job Spec Jerry, Resume Reggie, Ted and Thuli, the Crayon ATS, and semantic search on a platform with 175K+ active users, plus lead frontend work on The Bridge: the Drop, the 144-second Pause, The Crossing, Planks, The Locks, and The Collective Pulse.
02 — What I build
I build the product around the model. Frontend craft, the API, and the AI feature land together — from the requirement to something people can use.
01 — Frontend
React.js, Next.js, and TypeScript. Dashboards, authentication flows, role-based workflows, responsive layouts, and application state.
React.js · Next.js · TypeScript · JavaScript · Redux Toolkit · React Query · HTML · CSS · Tailwind · Material UI
02 — Backend
Node.js and Express. REST APIs, JWT authentication and authorization, MongoDB, third-party integrations, and AWS when the app has to stay up.
Node.js · Express.js · REST APIs · JWT · MongoDB · Mongoose · SQL · AWS
03 — AI
LLM integration, prompt engineering, structured outputs, RAG, embeddings, and vector search — shipped as product behavior, with the frontend and API around them.
Python · LLM APIs · Prompt engineering · RAG · Embeddings · Vector search · AI workflows
03 — Crayon · AI in the product
On Crayon, a person describes a role, uploads a CV, or sits an interview. The AI turns that into something the hiring workflow can use.
I worked on these experiences inside the live product — integrating them, not designing the whole platform. Open one for what it does, my part, and why it is in the product.
Someone describes a role in plain English. Jerry builds the job spec. They edit it until it is right, then post it. On the product, Jerry turns a plain-English hiring need into a usable role brief.
That spec is what gets posted and managed, so the AI output has to be editable and ready for the hiring flow.
I built this AI-driven experience: LLM workflows, prompt engineering, structured outputs, and the frontend and API path around them, then integrated it into production.
LLM APIs · Prompt engineering · Structured outputs · React · REST
A person uploads an existing CV. Reggie turns it into a structured Crayon profile they can use when applying, and that employers can find.
The profile replaces retyping the same history into every application.
I integrated Reggie into the production user flow: the interface, the API communication, and delivery through to a profile people actually use.
LLM integration · React · REST APIs · Production workflows
On Crayon, candidates browse jobs from their skills and preferences. Employers reach an existing candidate network instead of starting every search from zero.
Finding a role or a person has to work inside that network, not as a separate search demo.
I worked with RAG, embeddings, vector search, and indexing concepts so retrieval inside these experiences can use context, not only an exact keyword.
RAG · Embeddings · Vector search · Indexing · LLM integration
Where a role requires it, Ted conducts an AI-led screening interview. Thuli is the alternate AI interview experience. Candidates complete a Ted or Thuli interview on those roles.
The interview sits in the same flow as the application, so a person can track what happens next.
I worked on the AI video interview experiences and their path into production: the frontend, API communication, asynchronous processing, error states, and end-to-end delivery.
AI workflows · Async processing · React · REST APIs
The Crayon ATS is where applicants, stages, and communication live. Employers manage roles and pipelines there, alongside the candidate network and AI tools.
Jerry, Reggie, and the interviews only matter if the result lands in that workflow.
I worked on ATS management workflows and the AI-assisted parts of that flow, across the frontend and the APIs next to them.
AI workflows · REST · Role-based product flows · React · Node
04 — Selected work
Two production products. Crayon is an AI hiring platform at scale. The Bridge is an AI product experience I led on the frontend. Earlier roles are below.
Shipped on the platform
A hiring platform. Employers post roles, manage applicants in one ATS, and reach a candidate network. AI sits at specific points in that flow.
Every day a walker gets one Drop, steps into a 144-second Pause, crosses five Planks, and can leave one anonymous thought at The Locks.
The Drop is the daily question or provocation. It arrives at 13:44 local time. The walker opens The Bridge and reads it. There is one Drop a day, and no feed.
I led the frontend for how a walker meets that day’s Drop, including the public preview and the signed-in daily experience.
Next.js · React · TypeScript · Tailwind CSS
A private AI conversation with a hard stop at 144 seconds. The AI is a mirror, not a guru: it challenges and reflects, asks one question at a time, and does not simply agree. Input is capped at 144 characters. Responses stream. Each session uses that day’s editorial package with the Drop.
The conversation is stored without an identity. One Pause a day. It does not extend.
I led the frontend for the session: the timer, the conversation, and its place in the daily flow. The product calls Anthropic’s Claude through the API, with streaming. That sits in the wider architecture, beside the interface I built.
Next.js · TypeScript · Streaming conversation UI · 144-second cutoff
The Crossing is how a walker leaves The Bridge. Five Planks are curated links for that day. Before login they are a fixed set. After the Pause they are specific to that Drop. The product is built to send people onward, not to keep them in a feed.
I led the frontend for The Crossing and the Planks, including the difference between the public set and the Drop-specific set.
Next.js · React · TypeScript · Product flows
The Locks is the wall where a walker can leave one thought before they go. No names. No replies. The product does not store an identity with that thought.
I led the frontend for leaving a thought and for reading The Locks as a public, anonymous surface.
Next.js · React · Anonymous product flow
The Collective Pulse is an aggregate reading of anonymous Pauses. Four dimensions: Curiosity, Clarity, Anxiety, and Hope. It is never an individual score. Walkers are registered users. Stewards are practitioners connected through the product. There is also an admin side for operating the product.
I led the frontend for The Collective Pulse, walker functionality, steward-related surfaces, and the admin side of the product.
Next.js · TypeScript · Public product · Admin
Product architecture · not all of this is my implementation
Software Developer
Production web applications across the interface and the API. React.js, Node.js, Express.js, and MongoDB, with AWS in the deployment and troubleshooting path.
2023—2024
Freelance Developer
Freelance Developer
Full-stack web applications taken from client requirements through development and deployment.
05 — Experience
Client projects, production applications, then AI features on Crayon and The Bridge.
2025
May 2025 — PresentSenior Software Developer, AI / Full Stack · Remote · Freelance
2024
June 2024 — May 2025Software Developer · Indore, India
2023
Mar 2023 — Jan 2024Freelance Developer · Remote
2019
2019 — 2023B.Tech, Computer Science · CGPA 8.9 / 10
06 — Technical stack
Grouped the way the work is grouped. Every technology here is already on the page — nothing is hidden behind a hover.
07 — About
Indore, India. This is how a feature gets from a requirement to something people use in production.
Experience
3+
Years building software. Client projects and production applications, then AI features on Crayon and lead frontend work on The Bridge.
Now
Crayon Jobs Platform · Remote · Freelance
May 2025 — Present · Indore, India
Production scale
175K+
Active users on Crayon Jobs. Production features across the interface, the APIs, and the AI inside it.
01
AI capabilities start as a product need. I take them through a user experience and a backend integration, not a notebook that never ships.
02
Frontend flows, API communication, asynchronous processing, and error states are part of the same feature.
03
On Crayon that is a job spec, a profile, an interview, or an ATS step. On The Bridge it is the 144-second Pause.
04
Debug, integrate, and keep it responsive. I do that with product, design, and engineering — performance and reliability included.
The interesting part is not the prompt. It is the feature still working once real users show up.
08 — Contact
Full-stack AI engineering. Write to me — I read what I get.