Experience
Internships and paid engineering work.
NeoSapien
Jun 2026 – present
AI Engineer Intern · Remote / Bengaluru
- Speech-to-text evaluation for a shipping consumer wearable: harnesses that score transcription on hard, accented, code-switched Indian audio using word error rate, character error rate and entity recall together, because a single aggregate score hides exactly the failures that matter.
- Characterised failures by acoustic condition rather than ranking vendors, so a result answers what to change rather than only which option scored best.
- Wake-word evaluation across candidate trigger phrases, measuring detection rate and false-fire rate per phrase against a fixed recording set rather than on ad-hoc clips.
- Firmware-side test tooling for the device, including a USB command channel for microphone, tap and radio control so hardware behaviour can be exercised without a phone in the loop.
Dept. of Labour, Govt. of Madhya Pradesh
Nov 2025 – Feb 2026
System Analyst Intern · Indore, onsite · Client: MPSEDC
- Built e-SAHAYAK, a full-stack ecosystem in Flutter and TypeScript managing employer and worker relations and rights awareness.
- Found and fixed critical wage-tracking flaws in the attendance modules during UAT, before the mobile launch.
- Audited data parity across the staging to production deployment for Sambal 2.0, on a platform serving millions.
CytoScan Intelligence Technologies
Sept – Nov 2025
Data Infrastructure Intern · Remote / Chennai
- Owned the preprocessing and quality-control pipeline for BraTS 2023 brain-MRI data across 1,250+ patients, the stage that decides whether anything downstream is trustworthy.
- Voxel intensity normalisation across scans, because MRI values are not absolute: the same tissue reads differently between scanners and sessions, so a model trained on raw intensities learns the scanner rather than the pathology.
- nibabel-based loaders and visualisation tooling so a bad volume could be seen rather than inferred from a loss curve, which is the difference between finding a data fault in minutes and finding it after a training run.
- Benchmarked the resulting pipeline against published state-of-the-art results, and cut training latency.
Heizen (formerly OpenGig)
Aug – Sept 2025
Technical Program Manager Intern · Remote · Titan Capital-backed
- Translated product requirements into technical workflows and execution plans for 3 enterprise clients, working from briefs that were still changing while the work was being scoped.
- Built 0-to-1 prototypes to settle open questions early, so a disagreement about what the product should do was resolved against something runnable rather than against a document.
- Ran sprint and release coordination across those engagements at 90%+ on-time delivery.
ABV-IIITM Gwalior · C3iHub, IIT Kanpur
May – Aug 2025
Research Intern, Federated Learning for Healthcare · Gwalior, onsite
- Modular federated-learning pipeline over chest X-rays and electronic health records for pneumonia detection, with DenseNet and EfficientNet backbones swappable without touching the federation logic.
- The point of federating it is that patient data never leaves the institution that holds it: only model updates move, so the privacy constraint is structural rather than a policy someone has to honour.
- Hyperledger Fabric audit layer recording every model update, so a federation round cannot be altered after the fact without the record disagreeing.
- 94.2% AUC on non-IID data, which is the hard case: each site's data is distributed differently, so a result on evenly split data would not have told us anything about deployment.
Kanthariya Technologies
Dec 2024 – Apr 2025
Software Engineer Intern · Indore, onsite
- CMS platform for the DAVV University incubation centre, 100+ startups, 10K+ monthly users.
- GA4 analytics, SEO metadata and lead funnels, doubling onboarding engagement.
BuilderY
2025
Engineering Intern
CyberShield
2025
Digital Forensics Intern · offer via CIIS hackathon
Roomzy
2024
Developer Intern
E-Notebook
2024
Developer Intern · SGSITS
Research
Blockchain-Secured Federated Learning for Multi-Modal Healthcare
Privacy-preserving pneumonia detection. Presented at the IIITM Gwalior research symposium.
Research paper, second work
In progress.
Kaggle dataset
Published dataset.
About
I'm on a mechanical engineering degree and I've spent it writing software. That started as stubbornness and turned into the useful part. I don't have a lane I'm defending, so I go wherever the interesting problem is, Rust fuzzing infrastructure one month, a payments race condition the next.
What I actually care about is building scalable systems from scratch. That has meant government platforms on the Play Store, a provident-fund system moving over a crore a month, client backends in production, and two Chrome extensions people installed without being asked to.
For a while I contributed to open source the wide way, pull requests across a dozen projects at once. It taught me a lot and landed little. So I narrowed. Since January I've been in rust-lightning, on fuzzing and CI infrastructure, where the review bar is high and the code ships to people running Bitcoin nodes.
Shri G. S. Institute of Technology & Science, Indore
One of India's top GFTIs
B.Tech Mechanical Engineering, Minor in Machine Learning
CGPA 7.75, Minor 8.00
Sept 2023 – July 2027
Résumé
Read it here, or take the PDF.
Get in touch
Goes straight to my inbox.