Aagam Sheth

Aagam Sheth

I'm an M.Tech student in Computer Science and Engineering at Ahmedabad University. My work sits at the intersection of computer vision, large language models, and document intelligence. I am drawn to problems that require a model to reason about structure and layout rather than treating pixels or text as flat, undifferentiated input. This interest extends naturally into vision language models more broadly, and into information science as a way of understanding how knowledge is organized, represented, and retrieved.

Alongside my research, I serve as Chief Technology Officer at Tirth Infotech, where I direct technical strategy and oversee the delivery of secure production platforms for government and enterprise clients. The role keeps me equally engaged with systems design and applied research, and each informs the other.

I enjoy conversations that wander from research to everyday things, so feel free to reach out if you would like to talk shop over coffee or compare notes on a good cycling route. I can be reached at my email below.

contact [dot] me [at] aagamsheth [dot] com|CV|GitHub|LeetCode|Kaggle

Education

Ahmedabad University — 2025 - Present
M.Tech in Computer Science and Engineering (Data Science & Analysis and Artificial Intelligence)
School of Engineering and Applied Science · Advised by JB Sir · CGPA 3.45/4.00 (in progress)
Coursework: Deep Learning, Computer Vision, Large Scale Cloud Computing, Research Methodology
New L. J. Institute of Engineering and Technology — 2021 - 2024
B.E. in Computer Science and Engineering (Artificial Intelligence and Machine Learning)
Gujarat Technological University · CGPA 9.71
Coursework: Advanced Data Structures, Artificial Intelligence, Machine Learning, Computer Vision, Big Data Analytics
Government Polytechnic, Ahmedabad — 2018 - 2021
HSC (Diploma in Information Technology)
Gujarat Technological University · CGPA 9.76
Coursework: Programming Fundamentals, Web Technologies, Database Management, Operating Systems, Computer Networks
Ahmedabad International School — 2018
Gujarat Secondary Education Board · Percentile 96.5%
GSEB curriculum, Grades 7-10
Ahmedabad International School — Grades 1-5
Primary Years Programme (PYP) - International Baccalaureate (IB)

Publications

Label-Efficient Sclera Segmentation with Foundation Models: SSBC 2026 teaser
AI4Life Open Calls and Public Challenges: why, how, and what we have learned. teaser

Research Projects

Improving Online Trackers via Classical ML Feature Evolution
Small Object Detection & Tracking for UAV Videos
Unsupervised Denoising for Fluorescence Microscopy
Early Sepsis Detection from Electronic Health Records
LayoutLM vs. RoBERTa for Document Understanding
Reflex-Based Control & A* Planning for a Planetary Rover

Engineering Projects

Work Experience

Tirth Infotech
Chief Technology Officer · 2026 - Present
Technology strategy & delivery leadership
Own the technical direction of every Tirth Infotech engagement end to end, from bid-stage architecture and estimation through security reviews, deployment sign-off, and post-launch support. Set infrastructure standards for AWS and Gujarat State Data Centre hosting patterns, DMZ topologies, and CI conventions; choose stacks per project constraint rather than fashion; and mentor the engineering team through design and code review. Still hands-on: I review critical diffs personally and step into thorny performance or security problems directly.
Systems Architect & Full Stack Developer · Jan 2023 - Aug 2025
Archives Portal — Gujarat State Archives & Food and Drugs Control Administration
Architected and shipped the Archives Portal for the Gujarat State Archives and the Food and Drugs Control Administration, hosted on the Gujarat State Data Centre across three servers in a DMZ topology — one machine each for the Django/DRF application, PostgreSQL, and MongoDB. Built dynamic, metadata-driven APIs and admin panels so any archival schema can be modelled without code changes, added Gujarati full-text search over TsVector lexemes, RFC 7233 ranged downloads for large documents, and hash-table caching on hot client and server paths. The platform cleared two security audits by a CERT-In empanelled auditor and remains in production.
Full Stack Developer · Jul 2024 - Mar 2025
SahajPranali — Collector Office, Buldhana (Government of Maharashtra)
Built SahajPranali for the Collector Office in Buldhana, digitising land-revenue records — 7/12 extracts, mutation entries, land-acquisition files, and registers — behind search by district, taluka, village, and case number. Designed the secure dual-login model: OTP verification for citizens alongside role-based access for officials, plus watermarking and audit logging on every document view and download. Paid online downloads cut office visits and physical record handling, speeding up citizen services while preserving the legacy archive.
Developer · May 2022
Document Retrieval Portal — GSHSEB
Shipped in weeks, not months. A Karobari document retrieval portal for GSHSEB built on a static-generation model I chose after analysing how often the client's metadata actually changed — rarely — so static pages plus an AJAX search-suggestion panel beat a heavier dynamic stack on cost, speed, and reliability. Deployed solo-instance on AWS LightSail with Django and PostgreSQL.
Developer · Mar 2022 - Present
Objection Raising System — State Recruitment Boards
Recruitment exams attract objections by the thousand, and this platform is where candidates file them and boards resolve them. I developed the Django + MS SQL Server system deployed on AWS behind Apache Httpd and mod_wsgi, then made it fast: profiling showed a cryptographic hash burning CPU on every request, so a custom XOR integrity check replaced it — roughly 70% CPU saved across all eight cores and headroom for about 5,000 requests per minute on a single instance. Hot queries moved to Transact-SQL stored procedures, Django was hardened for thread safety under Apache MPM Prefork (10 processes x 100 threads), and I still maintain the system today.
Developer · Dec 2021 - Aug 2022
Stock Management System — GSHSEB
Custom stock management and tracking for GSHSEB covering purchase orders, sales orders, and internal and external transport movements across warehouses. Designed the ACID-compliant schema that kept accounting modules consistent, built more than ten reports shaped around how each office actually works, and gave administrators spreadsheet bulk-upload plus user management so operations never queued on developers. Django + PostgreSQL on AWS LightSail behind Apache.
Intern & Junior Developer · Jun 2018 - Aug 2021
Where it all started: five summers and part-time years maintaining VB.NET and C# WinForms applications for real clients with real deadlines. Customised over fifty Crystal Reports to match exactly what each office wanted on paper, travelled to client sites for on-premise deployments and support, and learned early the discipline of shipping software other people depend on.

Skills

Languages
PythonC/C++C#VB.NETJavaScriptSQLHTML/CSS
ML / Deep Learning & Computer Vision
TensorFlow/KerasPyTorchScikit-LearnHugging Face TransformersUltralytics YOLOSAHI (tiling inference)OpenCVNoise2Noise / Noise2VoidPandasNumPyMatplotlibSeabornOptunaSHAP
Web & Backend Frameworks
DjangoDjango REST FrameworkNext.jsReact.jsjQuery.NET FrameworkTailwind CSSshadcn/uiDataTables.jsChart.js
Databases & Messaging
PostgreSQLMySQLMS SQL ServerMongoDBRedisRabbitMQ
DevOps & Servers
AWS (Elastic Services, RDS, SageMaker, LightSail)DockerNginxApache httpd / mod_wsgiCron / CeleryPrometheus / Grafana
Techniques
PEFT / LoRA fine-tuningMulti-object tracking (BoT-SORT, Kalman)Classical descriptors (SIFT, HoG, LBP)Feature engineering & SMOTEFull-text search (TsVector)REST API designT-SQL stored procedures
Tools & Platforms
Git / GitHub / GiteaLaTeXJupyterLabPower BIPostman / InsomniaFigmaCrystal ReportsUbuntu / RHEL / Windows Server

Other

Outside of research and work:

ReadingPhotographyGuitar & other instrumentsCycling — happy to swap good routes around AhmedabadBrewing coffee — always up for talking pour-over ratios

Certificates

Mar 2023 Data Science Tools IBM Developer Skills Network [DS0105EN]
Mar 2023 Advanced SQL for Data Scientists LinkedIn Learning
Mar 2023 SQL for Data Analysis LinkedIn Learning
Mar 2023 Data Strategy NASBA & Project Management Institute (CPE Credit: 1.6)
Mar 2023 Python Data Analysis NASBA (CPE Credit: 4.8)
Mar 2023 Designing Highly Scalable and Highly Available SQL Databases LinkedIn Learning
Mar 2023 Machine Learning with Python: Foundations LinkedIn Learning
Mar 2023 Level Up: Advanced Python LinkedIn Learning
Feb 2023 Data Science 101 IBM Developer Skills Network [DS0101EN]
Feb 2023 Data Visualization with Python IBM Developer Skills Network [DV0101EN]
Feb 2023 Machine Learning with Python IBM Developer Skills Network [ML0101EN]
Feb 2023 Python 101 for Data Science IBM Developer Skills Network [PY0101EN]
Sep 2020 Introduction to Data Science in Python University of Michigan (Coursera)
May 2020 Programming for Everybody (Getting Started with Python) University of Michigan (Coursera)
Apr 2020 Python Data Structures University of Michigan (Coursera)
Jan 2013 GESE Grade 2 (CEFR A1) Trinity College London