↑
AJAY.

ajay mukund —engineer

I build things I'm curious about.

Mostly software. Sometimes a system. Occasionally things that shouldn't be as complicated as they are.

software · systems · architecture · AI-assisted building
Associate Engineer – Trainee @ Litmus7
local time--:--:--IST · Thiruvananthapuram
open projects ↓find me on linkedin ↗
Ajay Mukund A smiling.

ajay, mostly debugged

inspect the system ↗view source code ↗
01how i build

I follow problems. The domain is wherever the problem lives.

I tend to start with the thing that interests me, then figure out what needs to exist for it to work. That's taken me through security tooling, queue state machines, scheduling research, and an investigation OS. I don't plan for those areas — each one started with something I couldn't stop thinking about.

I use whatever makes sense. Sometimes that's code. Sometimes it's architecture first. Sometimes it means spending a week understanding a paper before writing a line. The tool choice is usually downstream of the problem.

The loop: problem → understand → design → build → iterate. Then the problem shifts slightly, and you start again.
this is actually how it goes
02selected work

Things I built far enough to understand the system underneath.

A few prototypes and a research paper. Some of these are finished. Some are still being figured out — I've tried to be honest about which is which.

honest caveat: these are prototypes, not production systems
01AI-assisted investigation operating system
Architecture V1 · feature-frozen

AnweshanaSahayi

"Investigators need shared, traceable state — not another chat window."

An investigation operating system for synthesizing fragmented digital evidence into a shared investigation state. I designed and implemented the architecture and prototype end-to-end, including the deterministic investigation backbone, reasoning/orchestration layer, data model, provenance, coverage, contextual graph projection, and investigator-facing workflow.

The backbone constructs factual investigation state with zero LLM calls. The reasoning agent reads structured state — not raw evidence — to identify the greatest remaining uncertainty and recommend the next inquiry. The investigator remains the final decision-maker.

architecture sketch · the investigation loop
↓↓↓↓↓↓↓

Hover or focus a stage to inspect its boundary.

FastAPIReact + TypeScriptPostgreSQL + pgvectorReact Flowprovenancehuman-in-the-loop

architecture decisionThe graph is a deterministic projection of investigation state, not the source of truth.

guardrailAI-derived relationships retain separate reliability and verification status; they do not become confirmed facts automatically.

statusArchitecture V1 frozen. Current focus: documentation, reproducibility, testing, and demo reliability.

Architecture V1 / testing, reproducibility, and demo readinessview repository — AnweshanaSahayi ↗
02real-time OP queue prototype
May 2026 — present

MediQ

"Hospital waiting shouldn’t feel confusing and uncertain."

A React/Vite and Firebase prototype for a real-time OP queue covering secure patient onboarding, appointments, walk-ins, live queue movement, tracker links, and ETA logic. Its interesting problem is state: holding pool versus active pool, physical presence, and doctor/reception/nurse workflows all affect what the queue means.

state sketch · a queue is a state machine pretending to be a UI
→→→→→

Hover or focus a pool or state to inspect the queue model.

React + ViteFirebase AuthFirestoreRBAClive ETA

architecture decisionQueue state follows real clinic workflows and physical presence rather than treating every appointment as an identical list item.

prototype / system projectview repo — MediQ ↗
03vehicular fog computing research
Sep 2025 — Apr 2026

SHybridQIGA — Vehicular Fog Scheduler

"How should scheduling behave when the workload keeps moving?"

A hybrid quantum-inspired genetic algorithm with heuristic seeding and deterministic, load-aware repair for multi-objective task scheduling in vehicular fog computing. The research evaluates static workloads and continuous SUMO-driven mobility through EdgeSimPy, including SLA, energy, cost, and latency trade-offs.

research sketch · scheduling under movement
latencyenergy
Pareto front candidate space

Illustrative only — no numerical values or experiment claims.

Hover or focus a point to inspect the trade-off idea.

PythonEdgeSimPySUMOmulti-objective optimization

evaluation noteUnder continuous mobility, the paper reports 38–58% fewer missed deadlines than GA across four topologies, while documenting an energy and cost premium against lightweight heuristics. It does not win every metric.

research paper / static workloads + continuous mobility evaluationview repo — Fog-Orchestrator ↗
04security and networking
CIAL internship · June 2025

DNS Spoofing Detection & Alert System

"Public Wi-Fi trusts a lot more than it should."

Built during my internship at Cochin International Airport. The Python/Scapy tool performs stateful DNS inspection, trusted-record and IP checks, TTL/metadata analysis, and transaction-mismatch detection, then surfaces events through a Flask-SocketIO dashboard and Telegram alerts with SQLite history.

security sketch · packet to alert
↓↓
vs
↓

Hover or focus a stage to inspect the trust boundary.

conceptual explanation · not a live network simulation
PythonScapyFlask + Socket.IOSQLite + SQLAlchemyTelegram alerts

security boundaryA controlled detection project focused on making quiet network trust visible and inspectable; the repository explicitly positions it for learning, experimentation, and authorized testing.

simulated detection project / internship workview repo — DnsCheck ↗
05browser extension
personal project

Haajar (ഹാജർ)

"Students shouldn’t have to make attendance decisions blindly."

A privacy-first Chrome extension for the RSET student portal that reads raw absence data, maps it against a batch timetable, and turns attendance into a decision. It uses no backend or external API calls; calculations and settings stay in chrome.storage.local.

Chrome ExtensionRSMS scrapingCSV timetable mappinglocal-only storage

architecture decision

supporting projectview repo — Haajar ↗
supporting work from the shelf —
PlacementGuardianChrome extension + FastAPI backend automating university placement drivesVoterFindDecodes legacy ISM fonts to make Kerala’s 2002 electoral rolls searchableTableDineQR-menu ordering system for a local biriyani shop
see everything on github ↗
03professional experiencefield log

Where I’m building now, and where I started.

2026Aug 10 → present
currently building

Associate Engineer – Trainee

Litmus7

Ten days in. Still figuring out the codebase, the team, and what I don't know yet. Interested to see what the next few months look like.

2025June · 1 month
internship

Project Intern · Cochin International Airport Limited (CIAL)

Worked with the enterprise security team on a simulated DNS-spoofing detection project. Got pulled deep into DNS internals, packet-level networking, and defensive tooling — the “DNS Spoofing Detection” project up above came out of this.

2024July · 1 month
internship

Bornbix Technologies

Data annotation and website work — one of my first hands-on stretches, and where I learned that the small details usually decide how something actually feels to use.

tools / focus

What I actually use. What I'm comfortable in.

Organised around how I actually spend time — not pretending the whole stack is equally important.

building

where most of my project time goes

exploring

  • agentic systems
  • investigation architecture
  • research prototyping
  • optimization
  • EdgeSimPy + SUMO

things I follow because the problems are interesting

background

grounding that shows up in the work

education & certifications

Where I studied. What I kept.

certifications
  • NPTEL — Cryptography and Network Security
  • Google — Foundations of Cybersecurity
  • NPTEL — Data Science for Engineers
04back cover / contact
if something caught your eye —

I'm most interested in problems where the system is still being figured out.

If you have something interesting — an architecture question, a weird problem, a project that keeps not quite working — I'm usually up for talking about it. Also open to roles, collaborations, or just a good conversation.

i check my email daily, usually
resume
Download PDFprofessional profile↗
github
AJAYM03github.com/AJAYM03↗
linkedin
/ajay-mukund03professional profile↗
medium
@ajayma92 published blogs↗
email
ajayma9@gmail.comopen a conversation↗