Hello, I'm
Mechanical Engineering Student | Fire Protection · Manufacturing · Robotics
Boston, MA
I'm a Mechanical Engineering student at Wentworth Institute of Technology in Boston. During an internship at a fire protection manufacturer, I helped design a fire suppression and leak detection system; the parent company filed a patent on the design. I have run 50+ pressure, leak, and reliability tests to IS, UL, and FM standards across two fire protection manufacturing roles and deployed suppression systems at data centers across India.
For my Senior Capstone, our 6-person team built X-FLEX, a compact scissor lift that moves loads through standard indoor doorways where full-size lifts don't fit. I worked on the Arduino-based control development. It's remote-controlled, drives on a chain-and-sprocket system with wheelchair DC motors, and lifts with a single linear actuator instead of hydraulics. All coursework complete as of Summer 2026; currently seeking a Spring 2027 Co-op in mechanical, fire protection, manufacturing, or product development.
Programming
CAD & Design
Manufacturing
Tools & Methods
Aug 2021 – Aug 2027
Wentworth Institute of Technology — Boston, MA
Aug 2018 – May 2020
RBK International Academy — Mumbai, India
Aug 2026 – Dec 2026 · Salem, NH
Moveras, LLC
Jul 2024 – Dec 2024 · Mumbai, India
Eurotech Cylinders Pvt. Ltd
Jul 2022 – Sep 2022 · Mumbai, India
Edelweiss Alternative Asset Advisors Limited
Jun 2022 – Aug 2022 · Remote
Inchtone
May 2022 – Jul 2022 · Mumbai, India
Corizo
Jul 2020 – Aug 2021 · Mumbai, India
Nitin Fire Protection Industries Ltd
Mar 2020 – Jun 2020 · Mumbai, India
Nitin Fire Protection Industries Ltd
Aug 2019 – Sep 2019 · Mumbai, India
DroneIT
Designed a fire suppression system with heat-based leak detection; completed full assembly model and manufacturing drawings in SolidWorks. The system integrates suppression and detection mechanisms to reduce response time and improve process safety.
Result: Patent successfully filed (NDA).
Senior Capstone project, 6-member team. Original concept used tank tracks, a hydraulic pump, and a PLC controller; reworked each as the design matured. Worked on embedded control with the team: programmed an Arduino Mega 2560 to parse FlySky RC receiver input and output PWM to Cytron MDD10A motor drivers for chain-and-sprocket differential drive. Lift motion comes from a single 24V linear actuator in place of the original hydraulic pump, with deadman and limit switches wired for fail-safe shutdown. Built to move loads through standard indoor doorways and elevators where full-size scissor lifts can't fit.
Result: Lifted 200 lb to 8 ft, held a 180 lb person jumping on the base plate with no damage, and cleared standard interior doorways and elevators at 34 in wide. Presented at Wentworth's Senior Capstone Showcase.
Modeled and assembled a high-speed wobbler mechanism via manufacturing, sand casting, and tumbling. Machined every part to a toleranced drawing set (ASME Y14.5-2009) and wrote per-part manufacturing reports to achieve stable operation at high RPM.
Result: 2,340 RPM achieved.
Machinists look up limits and fits every week, and the web calculators are covered in ads. pressfit answers in one command: hole and shaft limits, the clearance band, and what the fit feels like at assembly. The test suite is built from published fit-table rows, and a public verification ledger invites other engineers to check the numbers against their own handbooks.
Result: Live on PyPI as pip install pressfit; 34 tests, 23 of them published table rows; zero dependencies.
Paper submitted to ASTFE TFEC 2027, modeling heat traveling from a running engine's exhaust manifold, through a thermal-barrier coating, to the outer wall, then asking whether that surface gets hot enough to ignite spilled fuel with no spark. The dashboard reruns the paper's steady-state solver, RK4 soak-decay integrator, and grid-convergence study live in the browser.
Result: Interactive dashboard reproduces the paper's steady-state figures and 311–450°C autoignition band, computed client-side on every slider move.
Accepted paper (ICACR 2026, IEEE Xplore) proving a six-axis ABB IRB 120 arm can physically reach the poses a combined camera, depth-sensor, and laser-probe package would require for in-line inspection. The page renders a live 3D robot from the paper's real DH parameters and runs the same damped-least-squares inverse-kinematics solver in the browser.
Result: Live solver converges to the target inspection poses from the paper's reachability analysis, animated in 3D with a real-time workspace envelope.
Fuel consumption prediction where the model is not allowed to contradict the machine. A plain LightGBM happily predicts a heavier car burning less fuel between training points; guzzler adds monotone constraints so it cannot, and pays about 6 percent accuracy for the guarantee. Its worst miss traces to the Mazda RX-3, a rotary engine that breaks the displacement assumption, which is the kind of answer metrics alone never give.
Result: Beats the linear baseline on 18 of 20 splits, evaluated honestly with the noise stated.
Weekly demand forecasting for a real retailer's top SKUs, scored in the unit a stockroom pays for: safety stock at 95 percent service. A 12-week rolling backtest against the baselines every ops team already uses. The honest finding stayed in the README: the 4-week moving average beat LightGBM on accuracy, and measuring the error tails instead of assuming a textbook factor changed the shelf ranking again.
Result: Choosing the forecast well frees about 4,500 units of safety stock across 24 SKUs at the same service level.
Team entry from WIT's ASME chapter in the 2025-2026 IAM3D R.O.V.E.R challenge: a remote-operated rover that gathers sand and construction resources from an 8 ft by 8 ft course and deposits them into a bucket under first-person-view control, with the whole vehicle starting inside a 30.48 cm cube. As one of the seniors on a team of sophomores and juniors, I floated across the movement, manipulator, and frame subsystems, helping the underclassmen where they needed it; scoring weights the fraction of parts that are additively manufactured, so most of the body was 3D-printed.
Result: The team placed 5th in the ASME IAM3D R.O.V.E.R Competition.
9 AI-powered web apps built and deployed in 30 days. ProcureAI and IncidentIQ target engineering and manufacturing workflows directly, and Mechanical Vibrations Prep runs a live spring-mass-damper simulation alongside an AI tutor. All apps use GPT-4o, the Vercel AI SDK, and streaming structured output.
Result: All 9 apps live on Vercel. Built using OpenAI Codex as the primary coding agent; engineering decisions on scope and domain fit were mine.
Results pulled straight from the projects above.
I'm open to co-ops, research collaborations, and interesting engineering challenges. Can I design your next big thing? Feel free to reach out!
