Pratinav Seth — AI Researcher

TL/DR : I currently work as a Research Scientist at AryaXAI. I completed my B.Tech in Data Science from Manipal Institute of Technology, during which I worked at Mila, Bosch,IIT KGP and few startups. I was an 2023 AAAI Undergraduate Consortium Scholar and received MAHE UG Research Grant. I have reviewed and presented research at A*/A conferences such as AAAI, ICML, ACL & IJCNN and workshops at conferences such as NeurIPS, ICLR, MICCAI, AAAI, CVPR, EMNLP, WACV, MIDL, NAACL, EMNLP and ACL. ------- I currently work as a Research Scientist at AryaXAI (Arya.ai, an Aurionpro Company), where I work at the intersection of Explainable AI (XAI), AI alignment, and AI safety for high-stakes, real-world applications. My focus is on interpreting black-box models, evaluating XAI algorithm reliability, ensuring these systems are aligned and trustworthy, and exploring the use of foundation models for tabular data applications—especially in critical sectors. I recently completed my B.Tech in Data Science from Manipal Institute of Technology, during which I had the privilege of working at Mila Quebec AI Institute (under Dr. David Rolnick), Bosch Research India (with Dr. Amit Kale and Mr. Koustav Mullick), and KLIV Lab at IIT Kharagpur (PI: Dr. Debdoot Sheet). I conducted much of my research alongside my peers at Mars Rover Manipal AI Research (alongside Dr. Ujjwal Verma), Research Society MIT, and under Dr. Abhilash K. Pai . In 2023, I was honored to be selected as an AAAI Undergraduate Consortium Scholar, where I presented a proposal on Model Agnostic Uncertainty Aware Metrics. I have presented research at A*/A conferences such as AAAI, ICML & IJCNN and workshops such as NeurIPS, ICLR, MICCAI, AAAI, CVPR, EMNLP, WACV, NAACL, EMNLP and ACL. I have also served as a reviewer for various conferences and conference workshops such as NeurIPS, MICCAI, EMNLP, CVPR, ICLR, ICCV, ECCV and IJCNN. Passionate about applying artificial intelligence to solve real-world problems across various domains. I’m deeply passionate about building responsible AI systems that are aligned, safe, and transparent, with a particular interest in AI for Social Good and its applications in Medical Imagery and Remote Sensing. I’m always eager to connect and exchange ideas on AI research—let’s connect! 🚀 For a more detailed overview of my professional journey, projects, and contributions, feel free to take a look at my Resume.

Stackforce AI infers this person is a skilled AI Research Scientist specializing in Explainable AI and Deep Learning applications.

Location: Kolkata, West Bengal, India

Experience: 2 yrs 8 mos

Skills

Artificial Intelligence (ai)
Deep Learning
Explainable Ai (xai)
Leadership
Natural Language Processing (nlp)

Career Highlights

AAAI Undergraduate Consortium Scholar 2023
Presented research at top-tier AI conferences
Mentored over 10 undergraduate researchers

Work Experience

Lexsi Labs

Research Scientist (8 mos)

Research Scientist (11 mos)

Mila - Quebec Artificial Intelligence Institute

Researcher (5 mos)

Bosch Research

Computer Vision Research Engineer Intern (4 mos)

The Data Alchemists

Head of Artificial Intelligence (9 mos)

Research Society MIT Manipal

Co-President & Mentor AI Research (1 yr 1 mo)

Manipal Institute of Technology

Researcher (1 yr 3 mos)

Indian Institute of Technology, Kharagpur

Research Intern (1 yr 7 mos)

eedge.ai

Machine Learning Intern (2 mos)

Cureya

Data Science NLP Intern (1 mo)

Mars Rover Manipal

Researcher (2 yrs 6 mos)

Education

B.Tech at Manipal Institute of Technology

CBSE at Bharatiya Vidya Bhavan's

Pratinav Seth

AI Researcher

Kolkata, West Bengal, India2 yrs 8 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

AAAI Undergraduate Consortium Scholar 2023
Presented research at top-tier AI conferences
Mentored over 10 undergraduate researchers

Stackforce AI infers this person is a skilled AI Research Scientist specializing in Explainable AI and Deep Learning applications.

Contact

Skills

Core Skills

Artificial Intelligence (ai)Deep LearningExplainable Ai (xai)LeadershipNatural Language Processing (nlp)

Other Skills

API DevelopmentAnalytical SkillsArtificial Neural NetworksAutomotiveC++Cascading Style Sheets (CSS)Computer ScienceComputer VisionConvolutional Neural Networks (CNN)Data AnalysisData AnalyticsData CollectionData ScienceData VisualizationDeep Neural Networks (DNN)

About

Experience

Lexsi labs

2 roles

Research Scientist

Jul 2025 – Present · 8 mos

AI Research Scientist @ Lexsi Labs, Lexsi.ai
Tabular Foundation Models: Contributed to the development of foundation models for tabular data in high-stakes domains; co-developed a library for inference, fine-tuning, and benchmarking of tabular foundation models.
Interpretability-Guided Alignment: Investigating model optimization (pruning, quantization) and alignment (fine-tuning, RL-based alignment, unlearning) strategies across various model architectures—leveraging interpretability as a design principle and guiding mechanism.
Research & POCs: Led proof-of-concept (POC) projects for model optimization, fine-tuning, alignment, and internal research tooling to accelerate experimental workflows.
Leadership & Mentorship: Mentored six + research interns; led recruitment of interns and full-time scientists (Paris and India); authored technical and research documentation for stakeholders; initiated proof-of-concept projects to advance internal algorithmic capabilities.
Representation: Presented a poster at the MICCAI Workshop 2025.
Lexsi Labs, Lexsi.ai
Publications:
1. Interpretability-Aware Pruning for Efficient Medical Image Analysis. 2025. MICCAI Workshop 2025 (LNCS).
2. Interpretability as Alignment: Making Internal Understanding a Design Principle. 2025. Position Paper (Accepted at EurIPS Workshop on Private AI Governance).
3. TabTune: A Unified Library for Inference and Fine-Tuning Tabular Foundation Models. 2025. Technical Report — Open-Source Library.
4. Orion-MSP: Multi-Scale Sparse Attention for Tabular In-Context Learning. 2025. Pre-print.
5. Orion-BiX: Bi-Axial Meta-Learning for Tabular In-Context Learning. 2025. Pre-print.

Tabular Foundation ModelsInterpretability-Guided AlignmentResearch & POCsLeadership & MentorshipArtificial Intelligence (AI)Deep Learning

Research Scientist

Jul 2024 – Jun 2025 · 11 mos

AI Research Scientist @ AryaXAI Alignment Labs, AryaXAI.com
Research Focus: Working at the intersection of Explainable AI (XAI), AI alignment, and AI safety in high-stakes domains—interpreting black-box models, assessing XAI reliability, and developing foundation models for tabular data in fraud detection and mission-critical applications.
Explainability: Enhanced the DL-Backtrace method by generalizing its mechanics for model-agnostic use; co-developed a benchmarking framework for the systematic evaluation of XAI techniques.
XAI-Guided Optimization & Alignment: Investigating model-agnostic post-hoc optimization and alignment strategies across various model architectures—leveraging interpretability for safer, more reliable model behavior.
Leadership & Mentorship: Mentored two research interns; led recruitment of interns and full-time scientists (Paris and India); authored technical and research documentation for stakeholders; initiated proof-of-concept (POC) projects to advance internal algorithmic capabilities.
Representation: Served as R&D representative in client-facing engagements and presented AryaXAI solutions at industry forums, including the 5th MLOps Conference.
AryaXAI Alignment Labs, AryaXAI.com (Arya.ai — Lithasa Technology Private Limited, an Aurionpro Company).
Publications:
1. DL-Backtrace: A Model-Agnostic Explainability Method for Deep Learning Models. Accepted at IJCNN 2025.
2. XAI Evals: A Framework for Evaluating Post-Hoc Local Explanation Methods. Technical Report, 2025.
3. Bridging the Gap in XAI: Why Reliable Metrics Matter for Explainability and Compliance. Accepted at EurIPS Workshop on Private AI Governance, 2025

ExplainabilityXAI-Guided Optimization & AlignmentLeadership & MentorshipExplainable AI (XAI)Artificial Intelligence (AI)

Mila - quebec artificial intelligence institute

Researcher

Jan 2024 – Jun 2024 · 5 mos · Montreal, Quebec, Canada · Remote

Worked on a project (as a part of my Thesis) focused on computer vision and deep learning for geospatial applications targeting climate change, specifically in detecting abandoned oil and gas wells. Engaged with domain experts from McGill University to create a new geospatial dataset and benchmark deep learning models.
PI & Mentor: Dr. David Rolnick
Publication : Alberta Wells Dataset: Pinpointing Oil and Gas Wells from Satellite Imagery
Accepted at ICML 2025 and TCCAI Workshop, ICLR 2025 .

Computer VisionDeep LearningGeographic Information Systems (GIS)Artificial Intelligence (AI)

Bosch research

Computer Vision Research Engineer Intern

Jun 2023 – Oct 2023 · 4 mos · Bengaluru, Karnataka, India · On-site

Worked on a project focused on vision-based generative AI for autonomous driving, utilizing Latent Diffusion Models to generate additional data for difficult or misclassified samples. This approach aimed to improve the fine-tuning and optimization of the downstream task network.
Mentor: Kaustav Mullick (CR/RTC-IN)

Generative AIDeep LearningMachine LearningArtificial Intelligence (AI)

The data alchemists

Head of Artificial Intelligence

Oct 2022 – Jul 2023 · 9 mos

Co-founded and led a student club as the Head of Artificial Intelligence and Machine Learning, overseeing division activities and membership growth to over 30 individuals. Recruited initial members, conducted online workshops, and organized events. Guided club members in AI projects, focusing on machine learning, natural language processing, and computer vision.

Natural Language Processing (NLP)

Research society mit manipal

Co-President & Mentor AI Research

Aug 2022 – Sep 2023 · 1 yr 1 mo

Led and managed an undergraduate organization with over 90 members across more than 10 technical domains. Facilitated multidisciplinary collaboration and communication, recruited new members, and oversaw their learning and development. Raised awareness about the importance of undergraduate research and contributed to interdisciplinary projects, particularly in artificial intelligence. Mentored over 10 sophomore and junior undergraduates in various AI topics.
Publications :
1.Uncertainty-aware test-time augmented ensemble of berts for classification of common mental illnesses on social media posts.(Tiny Papers,ICLR 2023)
2.Sailing through spectra: Unveiling the potential of multi-spectral information in marine debris segmentation. (Tiny Papers, ICLR 2024)
3. Rsm-nlp at blp-2023 task 2: Bangla sentiment analysis using weighted and majority voted fine-tuned transformers. (Bangla Language Processing Workshop, ACL 2023)
4. Hgp-nlp at biolaysumm: Leveraging lora for lay summarization of biomedical research articles using seq2seq transformers. (BioNLP 2024 Workshop, ACL 2024)

LeadershipMentorship

Manipal institute of technology

Researcher

Jun 2022 – Sep 2023 · 1 yr 3 mos · Udupi, Karnataka, India · On-site

Under Dr. Abhilash K. Pai - Worked on skin-tone aware skin lesion classification using vision attention models and ExplainableAI techniques and currently working on Trustworthy AI solutions for Skin Lesions with support from MAHE UG Research Grant.
Publication: Does the Fairness of Your Pre-Training Hold Up? Examining the Influence of Pre-Training Techniques on Skin Tone Bias in Skin Lesion Classification
Accepted as Oral at PreTrain Workshop , WACV 2024.
Under Dr. Vidya Rao - Worked on an intersection of Cybersecurity and Artificial Intelligence involving Multi-class malware classification using various Deep Learning and Machine Learning techniques. Placed 5th among 134 Teams, 13th International Cyber Security Data Mining Competition 2022.

Natural Language Processing (NLP)

Indian institute of technology, kharagpur

Research Intern

May 2022 – Dec 2023 · 1 yr 7 mos · Remote

Worked extensively with chest radiographs, leveraging deep learning techniques to integrate domain knowledge using Graph Neural Networks (GNNs) into convolutional neural networks (CNNs) for hierarchical disease identification. Focused on enhancing model explainability.
Mentor: Rakshith Satish
PI: Dr. Debdoot Sheet

Deep LearningNatural Language Processing (NLP)Artificial Intelligence (AI)

Eedge.ai

Machine Learning Intern

Mar 2022 – May 2022 · 2 mos · Bengaluru, Karnataka, India · Remote

Worked directly with the CTO to develop a deep learning-backed intelligent platform. Utilized data analysis, machine learning, generative models, explainable AI, and advanced deep learning techniques to drive product innovation and development.

Natural Language Processing (NLP)Machine Learning

Cureya

Data Science NLP Intern

Jan 2022 – Feb 2022 · 1 mo · Noida, Uttar Pradesh, India · Remote

Developed a proof of concept pipeline for the conversational healthcare chatbot, Reyana, for various company products. This role involved applying natural language processing (NLP), natural language understanding (NLU), and natural language generation (NLG) techniques.
https://youtu.be/lAHyDV-uINU

Mars rover manipal

Researcher

Jun 2021 – Dec 2023 · 2 yrs 6 mos · Manipal · Hybrid

Progressed from a Student Trainee to a Senior Researcher, leading a student led independent team in the AI Research Wing. Contributed to projects in machine learning, deep learning, computer vision, and natural language processing, focusing on image-to-image translation, AI4SG, medical imagery, multi-modal AI, robust neural networks, and sentiment analysis. This work led to multiple publications at various A*/A conferences workshops. Also mentored over 10 undergraduates, with many collaborations resulting in publications in with some being the mentees' first publications.
Publications:
1. Lapgsr: Laplacian reconstructive network for guided thermal super-resolution. (Journal Paper under Review)
2. Lamar: Laplacian pyramid for multimodal adaptive super resolution. (Student Abstract and Poster, AAAI 2024)
3. Corefusion: Contrastive regularized fusion for guided thermal super-resolution. (CVPR Workshop 2023)
4. Refuseg: Regularized multi-modal fusion for precise brain tumour segmentation. (BrainLes Workshop, MICCAI 2023)
5. Uatta-ens: Uncertainty aware test time augmented ensemble for pirc diabetic retinopathy detection. (Medical Imagery Meets NeurIPS Workshop, NeurIPS 2022)
6. Evaluating predictive uncertainty and robustness to distributional shift using real world data. (Bayesian Deep Learning Workshop, NeurIPS 2021)
7. Agrillm: Harnessing transformers for farmer queries. (NLP for Positive Impact Workshop, EMNLP 2024)
8. Performance evaluation of deep segmentation models on landsat-8 imagery. (Tackling Climate Change with Machine Learning Workshop, NeurIPS 2022)
9. SSS at SemEval-2023 task 10: Explainable detection of online sexism using majority voted fine-tuned transformers. (SemEval Workshop Shared Task, ACL 2023)
10. Analyzing effects of fake training data on the performance of deep learning systems. Pre-Print.