Projects

Prototypes, systems work, and earlier projects.

A mix of prototype AI-assisted tools, exploratory systems work, and earlier academic and embedded systems projects. The common theme is end-to-end systems that real people can actually use.

Current focus

Right now I’m exploring AI-assisted tools and designing prototype software systems. Most of this work is personal experiments and internal tools; some ideas are intended for future products or services when the timing and status are right.

AI assistants & workflow tools

Building LLM-powered assistants that help people move through long, multi-step workflows with less stress and fewer mistakes. These are currently prototypes and reference designs, not public products.

  • Structured, typed data models instead of unbounded free-form notes.
  • Step-by-step flows with clear confirmations and “checkpoints.”
  • LLMs used for judgement, drafting, and summarization – not hidden magic.
  • Designed so users can always see what the system knows and why.
  • TypeScript
  • Node
  • PostgreSQL
  • LLM orchestration

Custom Software Prototypes

Small, opinionated tools and example apps built around real constraints: scheduling, planning, money, and everyday logistics. These are currently personal prototypes and reference implementations, not active client work.

  • Designing and implementing lightweight web apps and dashboards.
  • Exploring mobile-friendly UIs and flows for busy, non-technical users.
  • Putting reliability and debuggability ahead of shiny complexity.
  • React
  • FastAPI / Node
  • REST APIs

Personal tools & experiments

Internal tools to manage personal finances, planning, and recurring tasks — basically building the systems I wish I’d had earlier. These are private utilities I use myself, not commercial products.

  • Budgeting and cash-flow planning around real-world debt and 0% APR offers.
  • Simple automations to reduce repetitive admin work.
  • Iterating quickly on ideas before turning them into full products.
  • Python
  • Spreadsheets & scripts
  • Backend services

Academic projects

Earlier projects from my B.Tech and M.S. that built my foundation in embedded systems, distributed systems, and assistive technology.

ML-Based Attack Detection for Cooperative Adaptive Cruise Control

M.S. research project focused on connected vehicles and cooperative adaptive cruise control (CACC), exploring how to make vehicle platoons more robust against adversarial behavior and sensor faults.

  • Designed and simulated a CACC scenario with communicating vehicles under normal and attack conditions.
  • Implemented machine-learning based detection to identify malicious or faulty vehicles in the platoon.
  • Evaluated impact on safety and stability when isolating compromised nodes from the control loop.
  • Connected vehicles
  • CACC
  • Machine learning
  • MATLAB / Simulink

Twitter Clone – Distributed Systems in Elixir & Phoenix

Designed and implemented a working Twitter-style engine using Elixir and the Phoenix framework to explore distributed systems concepts.

  • Implemented core functionality of a microblogging platform in a distributed setting.
  • Focused on real-time updates and client–server communication.
  • Used the project to deepen understanding of fault tolerance and concurrency.
  • Elixir
  • Phoenix
  • Distributed Systems

Real-Time Embedded System for Differently Abled and Seniors

B.Tech final project: A portable single-seater electric vehicle with an embedded system and Android UI, designed to help differently abled and senior users move more safely and independently.

  • Developed an Android app UI to control speed, direction, and braking.
  • Monitored voltage, current, speed, and temperature in real time on the UI.
  • Implemented Bluetooth communication (HC-05) between the app and an MSP430-based embedded system.
  • Designed relay-based control for rear drive and steering, with safety and stability in mind.
  • MSP430
  • Android
  • Bluetooth (HC-05)
  • Embedded Systems

Embedded & IoT-style prototypes

Wheelie

Robotic Arm Equipped Solar Powered Hybrid Electric Wheelchair Assistive mobility prototype for the Texas Instruments Innovation Challenge India Design Contest 2015, aimed at giving wheelchair users more independence in everyday tasks.

  • Equipped with two 6-DOF robotic arms (≈1.2 m reach, ≈700 g payload) to handle tasks like reading, eating with a spoon, and pouring water.
  • Solar-powered hybrid electric wheelchair: on-board panel combined with a 2-DOF mechanical solar tracker and a Perturb & Observe MPPT algorithm to improve charging efficiency.
  • Head-tracking based wheelchair control and eye-tracking based end-effector control, implemented with image processing on a BeagleBone.
  • Inverse kinematics on the BeagleBone for coordinated arm motion and safe operation around the user.
  • Designed as a low-cost, eco-friendly alternative to existing electric wheelchairs to improve quality of life and accessibility.
  • Selected as a semi-finalist (top 86 out of 1209 teams) in the Texas Instruments Innovation Challenge India Design Contest 2015.
  • Embedded Systems
  • Robotics
  • Computer Vision
  • Solar Power
  • Assistive Tech

Smart Watch (under $30)

A prototype smartwatch that mirrors SMS and missed-call information from a phone, built entirely from low-cost components.

  • Developed an Android app to forward incoming messages and missed calls to the watch.
  • Used an MSP430 microcontroller and HC-05 Bluetooth module for communication.
  • Displayed messages and caller numbers on an LCD screen; implemented basic timekeeping.
  • Placed as second runner-up at Sangam Pragyan 2014 among 300+ teams.
  • MSP430
  • Android
  • Bluetooth
  • Embedded UI

Wireless Home Power Monitor

A non-invasive home power monitoring system with live measurements and visualization to help understand and manage energy usage.

  • Used non-invasive current sensors and voltage sensing to measure load and voltage.
  • Computed real-time voltage, current, power factor, and power on an MSP430.
  • Sent data over Bluetooth to an Android app for live visualization.
  • Used zero-crossing detection for power factor estimation; graphed power vs time and PF vs time.
  • Aimed at identifying spikes/surges and enabling smarter energy usage.
  • MSP430
  • Power Electronics
  • Android
  • Data Visualization

Theme

The through-line across these projects is systems thinking: understanding constraints, designing an architecture that respects them, and then building something that real people can use without needing to see all the complexity underneath.

Contact

If you’d like to know more about specific projects, reach out via LinkedIn.