Building software for messy real-world operations where generic SaaS fails.
From diagnostic lab benches to Indian D2C attribution and edge UAV avionics.
I'm a founder and builder who uses modern AI tooling to ship real software for messy real-world operations. My stack is whatever the problem actually needs — if an LLM or agent can write it, I can orchestrate it, test it against physical constraints, and ship it. Currently studying Data Science & AI (DSAI) at Masters' Union, between Rajkot and Delhi.
Currently building Kliniqo (live at kliniqo.co.in ↗ — an automated LIS + ASTM machine interface with a working MVP deployed in an active lab) and Ecommerce Hub (an analytics & reconciliation truth layer for Indian D2C brands, sparked by building SURGE Men at surgemen.in ↗). I don't get lost in syntax debates — I ship things that run.
Every serious project, past experiment, and honest failure. Click any card for the full breakdown with architecture diagrams, verified numbers, and real stories.
Diagnostic workflow & LIS software, live at kliniqo.co.in. Ingests ASTM E1394 serial data directly from haematology & biochemistry machines, auto-drafts reports in 3s, and automates patient recall. Tested on real family lab bench (Vaibhav Lab).
The public booking site for an 18-year-old Rajkot pathology lab, built on Kliniqo and live to real patients. Searchable test catalogue, cart and checkout, home-collection booking, WhatsApp reports, and the full interface in English, हिंदी and ગુજરાતી. This is the proof: go click it.
Complete high-performance men's hair styling brand live at surgemen.in — Matte Volumizing Texture Powder (shown rotating 360° below), alcohol-free matte clay, sea salt spray, curl cream, pomade, and styling mousse — formulated with cosmetic chemists. Scaled digital portal; physical production paused at 5,000+ unit MOQ.
A health app that connects to whatever you already wear — Fitbit, smart rings, watches — pulls that continuous signal, merges it with your actual diagnostic reports, and reasons over both together instead of leaving you with two disconnected sets of numbers. Your wearable knows your resting heart rate; your blood report knows your thyroid. Nothing today reads them as one person. MVP is being built right now.
Brand identity, product storytelling and the full marketing site for an NFC social card (Devil / Angel / Witness). I built the site and the brand — the product itself is not mine to run. Included because it shows the design and frontend range.
A per-order contribution-margin dashboard for Indian D2C brands, designed to close the 60%+ COD and 25-28% RTO return leakage by joining Shopify orders, Razorpay settlement, Shiprocket tracking and Meta/Google ad spend. Prototype built out of the SURGE post-mortem — not yet pointed at a live brand’s numbers.
Native 24/7 action-oriented AI assistant idea for booking, lead triage, and Web/WhatsApp automation. Researched in vault, verified horizontal commoditization risks, and made the disciplined call not to write unvalidated code.
Deterministic n8n workflow engine built for Vaibhav Laboratory. Automatically confirms test appointments, dispatches PDF test reports via WhatsApp when signed, and triggers quarterly diabetic/thyroid patient recalls. Zero AI slop — pure rock-solid automation.
Personal study tool built for fun. Ingests university pre-reads, lecture slides, and syllabus modules into an interactive concept graph. Extracts entity connections and spits out high-yield active-recall briefs before exams.
Rule-based momentum + RSI order execution script written on Zerodha Kite API. Built as a coding exploration — never live tested with real capital, never generated returns. Parked honestly.
Had 3 days to finish a crucial JEE chemistry chapter. Built a full burndown chart, streak tracker, and estimate calculator in React instead. Chemistry survived; barely.
Automated staking yields and epoch compounding engine on Solana. Ran smoothly in background without human intervention. Zero drama, exactly how bots should run.
Not shipped. Not started. These are the problems I would drop everything for — and the reason I keep sharpening on everything above.
Tactical edge compute, MAVLink telemetry, PX4 flight control, and autonomous obstacle navigation for UAVs. I have built the avionics simulator and the CAD model below. What I actually want to build is the real one — running on hardware, where a dropped packet is not a bug report.
Every AI tool still demands you learn to talk to it. The people who need it most — the lab technician at 8 PM, the warehouse kid reconciling returns, the shop owner with one laptop — never will, and no amount of onboarding changes that. I want to build software where the AI disappears into the job: no chat box, no prompt craft, no blank cursor. It watches the work and does the next obvious thing.
Interactive 3D CAD schematic for autonomous edge computing and companion hardware. Drag to orbit, inspect sub-systems, and explore fail-operational architecture.
A playable real-time tactical air defence simulator extracted directly from Priyam's Autonomous Defence AI companion avionics track. Real canvas physics, Kalman trajectory prediction, kinetic interceptors, drone swarms, and authentic telemetry.
Whether you want to discuss pre-seed funding & investment, automate something painful with agentic AI, or just drop by for pure unhinged bakchodi and banter — I'm in for everything. Email is fastest.
Kem chho! 🚀 Build mode ON!
