Abhishek
Creator of Unbrain — a production AI interview platform serving 1,000+ active university students — and published researcher in cryptography & steganography.
Building software that works
I'm a frontend engineer with 4+ years shipping React (Native + Web) and Next.js applications to production across fintech, logistics, and AI domains. I've delivered for 10+ clients and led small teams, owning full project lifecycles from architecture through deployment on AWS, Azure, and Vercel.
Along the way I founded Unbrain, a production AI interview platform serving 1,000+ active university students across 17 companies — and published research on AES-driven steganographic security.
“Clean architecture, clear communication, and code that ships.”
Technologies
Competencies
Where I've worked
Work History
Freelance Frontend Developer
Self-Employed · Remote (US, Europe, Gulf)
- Web and mobile frontends for 10+ clients across logistics, fintech, and e-commerce
- Customer-facing apps, internal dashboards, and admin portals for startups and established businesses
- Full project lifecycle ownership: requirements, architecture, implementation, API integration, deployment (AWS EC2/S3/Lightsail, Azure, Vercel)
- Async-first collaboration across US Eastern, European (CET), and Gulf (GST) time zones
Frontend Developer & Team Lead
Hefi Capital Ventures · Remote
- Architected a subscription-based cab platform with 3 interfaces — React Native rider app, React/Next.js ops dashboard, fleet-owner portal — serving 500+ riders with 99.5% uptime
- Integrated Paystack recurring billing end-to-end: payment flow, plan management, reconciliation dashboards
- Led a team of 4 developers: code review process, 2-week sprint cycles, technical direction aligned with product goals
- Collaborated with backend engineers on REST API design and translated stakeholder requirements into technical specifications
Commercial Projects
Unbrain — AI Mock Interview Platform
Live- Production AI mock-interview platform serving 1,000+ active university students across 17 companies (TCS, Amazon, Google, etc.)
- State-machine-driven flow with adaptive questioning; AWS Bedrock for interview generation with prompt caching and a pre-AI detection chain (gibberish, frustration, skip intents) that cut unnecessary API calls, latency, and cost
- Indian-English voice interviews via Sarvam AI TTS/STT with sentence-split prefetch pipeline and blob-based audio playback for hands-free conversation
- Admin panel with JWT auth, session monitoring, audit logs, and feature flags; 23-table PostgreSQL schema on AWS RDS with Prisma ORM; AI-generated learning roadmaps with graph visualization
Academic publication
Secure Random Pixel Distribution (SRPD) Algorithm
PublishedSecurity and Privacy · 2025
A novel cryptography-integrated steganography method for secure image data embedding — an AES-driven framework for pixel distribution achieving PSNR 70–85 dB and SSIM > 0.998 across benchmark images while maintaining linear computational complexity.
Key Contributions
- AES-driven cryptographic framework for image steganography via secure random pixel coordinate mapping
- PSNR 70–85 dB and SSIM > 0.998 across 10 benchmark images with linear computational complexity
- Classified "NOT DETECTED" by standard steganalytic tools (chi-square, RS analysis)
- Potential applications in digital forensics and secure communications
Interested in research collaboration or discussing this work?
What I've built
Unbrain
A production AI interview platform serving 1,000+ active university students across 17 companies including TCS, Amazon, and Google. A state-machine-driven flow conducts realistic spoken interviews with adaptive questioning, then delivers real-time scoring, gap analysis, and AI-generated learning roadmaps.
Key Features
- AWS Bedrock interview generation with prompt caching and a pre-AI detection chain (gibberish, frustration, skip intents) that reduced unnecessary API calls, latency, and cost
- Indian-English voice interviews via Sarvam AI TTS/STT with sentence-split prefetch pipeline and blob-based audio playback for hands-free conversation
- Full admin panel with JWT auth, session monitoring, audit logs, feature flags, and AI provider config
- 23-table PostgreSQL schema on AWS RDS with Prisma ORM; graph-visualized, AI-generated learning roadmaps
Subscription Cab Platform — Hefi Capital Ventures
A subscription-based multi-interface cab platform serving 500+ riders with 99.5% uptime: a React Native rider app, a React/Next.js operations dashboard, and a fleet-owner portal — built around monthly plans instead of per-trip payments.
Key Features
- Three interfaces: rider app, ops dashboard, fleet-owner portal
- End-to-end Paystack recurring billing: payment flow, plan management, reconciliation dashboards
- Real-time booking state management across user sessions
Let's work together
Whether you have a project in mind, an opportunity to discuss, or just want to say hello — I'd love to hear from you.
Send an email