AI Researcher

Sachish Singla

Computer Vision • Federated Learning • Reasoning Systems

Vision-Language Models Robust ML Data-Centric AI Systems

Integrated B.Tech + M.Tech student at IIT Kharagpur focused on building reliable learning systems. I work at the intersection of vision, representation learning, and federated settings, translating ideas into rigorous experiments and deployable pipelines.

Profile photo of Sachish Singla

IIT Kharagpur

Computer Science & Engineering

About

I'm a Computer Science student at IIT Kharagpur with a focus on AI research, particularly in vision and federated learning. I enjoy building end-to-end research prototypes, from data curation and modeling to evaluation and reporting.

I've collaborated with IBM Research, KAIST, and Preimage on projects spanning reasoning data, representation learning, and 3D scene understanding. I value rigorous baselines, reproducibility, and clear scientific communication.

Research Focus Robust ML + VLMs
Competitive Programming Codeforces Expert
Inter-IIT Multiple Gold Medals

Skills & Technologies

Python
C/C++
PyTorch
Computer Vision
Machine Learning
Git

Education

Indian Institute of Technology, Kharagpur

Integrated B.Tech + M.Tech (5Y) in Computer Science and Engineering

2022 - 2027

Pursuing integrated dual degree program in Computer Science and Engineering. Actively involved in research, competitive programming, and various technical competitions.

Experience

Summer Intern

IBM Research | Bengaluru, India

May 2025 - Jul 2025

Computer Vision Intern

Preimage | Bengaluru, India

Jan 2024 - Apr 2025

Visiting Student Researcher

KAIST | Seoul, South Korea

May 2024 - Jul 2024

Featured Projects

IBM Research: Reasoning Dataset Curation & Benchmarking

Built data curation workflow using FineMath, OpenWebMath, and Sangraha; automated Q&A pair generation for mathematical reasoning datasets; conducted empirical studies on test-time scaling behavior and performance trade-offs.

Python LLMs Data Engineering

Preimage: 3D Segmentation & Scene Understanding

Devised pipeline leveraging Grounded SAM and MaskClustering for 3D segmentation on 360° images; integrated multi-view correspondence with vision-language models to track scene evolution and generate detailed progress reports.

Computer Vision Grounded SAM VLMs

KAIST: Reward-Prediction Representation Learning for RL

Developed reward-prediction driven representation learning achieving up to 2x improvement in sample efficiency for RL tasks via LSTM-based temporal modeling; surveyed 40+ papers on action recognition and procedural understanding.

Reinforcement Learning LSTM Representation Learning

Mixup-VFL: Vertical Federated Learning

Formulated a novel label-mixing framework for Vertical Federated Learning, reducing RMSE by up to 72% for regression tasks with minimal data overlap. Published at IEEE ICDCSW 2025.

Python PyTorch Federated Learning

Adobe Image Classification | Inter-IIT Gold

Integrated EfficientNet-V2-S with Adversarial Training on Purification (ATOP) to classify low-resolution real and AI-generated images. Achieved 43.77% F1 score for artifact identification.

PyTorch EfficientNet CLIP

DevRev Tooling | Inter-IIT Gold

Built a dual-agent dataset generation pipeline via Self-Instruct framework to create 2,000+ query-output pairs for instruction finetuning. Achieved 87.79% JSON similarity on tool-augmented LLM tasks.

Python DeepSeek Self-Instruct

Food Delivery Service Management

Developed a MERN stack web application for food delivery platform with client-side interfaces, dynamic management dashboards, JWT authentication, and real-time order tracking.

MongoDB Express React Node.js

Reliable Transport Protocol

Implemented a custom reliable transport protocol over UDP with sliding window flow control for ordered and lossless packet delivery. Designed multi-threaded network system with shared memory.

C UDP Networking

Publications

2025

Mixup-VFL: Leveraging Unaligned Data for Enhanced Regression in Vertical Federated Learning

Sachish Singla, Prudhvi G., Ayush K., Devodita C., Varun T., Avi A., Debashish C.

IEEE ICDCSW 2025

Formulated a novel label-mixing framework for Vertical Federated Learning, reducing RMSE by up to 72% for regression tasks with minimal data overlap. This work addresses the challenge of data heterogeneity in federated learning scenarios where participants have different feature spaces.

Get In Touch

Let's Connect

I'm always interested in research collaborations, internships, and academic discussions. Feel free to reach out.

Email (Institutional) singla.sachish@kgpian.iitkgp.ac.in
Email (Personal) sachishs.15@gmail.com
IIT Kharagpur, India