You can download my full CV as a PDF by clicking the button below.
Quick Links
Navigate to different sections or explore related pages:
Summary
AI/ML Engineer with expertise in:
- Large Language Models (LLMs): Agent orchestration systems, evals, finetuning.
- Natural Language Processing: Text analysis, multilingual systems, conversational interfaces
- Speech Recognition: Voice assistants, ASR systems, acoustic modeling
- Cryptography & Security: Cryptanalysis of AES, authenticated encryption
Professional Experience
Ladduu AI
Co-Founder & CTO | New York, NY | April 2025 - Present
- Building agentic care for senior citizens by helping them with digital & physical tasks, social connection.
- Built multilingual voice assistant for iOS that can execute actions.
- Designed and implemented a WhatsApp assistant supporting over 8 Indian languages, providing the first taste of AI for thousands of Indians over age 45.
Skylink (Y-Combinator W22)
Founding Machine Learning (AI/ML) Engineer | New York, NY | July 2023 - May 2025
- Lead ML team of 6 in building AI-driven travel agent for corporate travel using Large Language Models (LLMs).
- Engineered LLM-based agent orchestration system driving $3M ARR through enterprise deployment (40k+ users).
- Spearheaded core AI development resulting in BTN Innovation Faceoff 2024 award.
Columbia University
Teaching Assistant (TA) | New York, NY | February 2022 - May 2023
- Courses: Fundamentals of Speech Recognition, Natural Language Processing, Programming for Social Impact.
- Mentored 100+ students in Kaldi, Tensorflow, PyTorch, PHP, HTML/CSS attaining 4.76/5 TA rating.
INfACT (Early-Stage Startup)
Tech Lead New York, NY | March 2022 - January 2023
- Partnered with Prof. André Corrêa d’Almeida in transforming initial idea to client-facing MVP in < 3 months.
- Single-handedly built analytics dashboard for raw LinkedIn data to monitor industry-specific postings using NLP techniques in Python, spaCy, Tableau. Recognized as Columbia Tech Ventures invention.
- Executed GTM strategy for NYC DOT, Niteroi City, NYC Mayor’s Office achieving > 95% client satisfaction rate.
Fusemachines Inc.
Machine Learning Intern, AI Services | New York, NY | June 2022 - August 2022
- Collaborated with team of 5 to build Data Science platform prototype leveraging Kubeflow, MLflow in 10 weeks.
- Implemented NLP module using Tensorflow, BERT to detect named entities in medical notes with 76% accuracy.
Ladduu (Early-Stage Startup)
Co-Founder & Chief Technology Officer | Andhra Pradesh, India | May 2021 - August 2022
- Led Product team of 7 in developing speech-enabled hyperlocal e-commerce apps for over 300 users.
- Engineered multi-lingual SaaS mobile/web apps (Kotlin, NodeJS, ReactJS) generating 6-figure ARR per customer.
de.ci.phe.red Lab, IIT Bhilai
External Collaborator (Information Security) | Bhilai, India | August 2020 - Present
- Mentored team of 4 in benchmarking lightweight authenticated encryption schemes for India’s 5G network project.
- Formulated content theft exploits on Indian OTT music platforms making Wynk Music update its security policies.
IIT Kharagpur
Junior Research Fellow | Kharagpur, India | October 2019 - May 2021
- Researched new impossible differential attacks on reduced round AES cipher for higher number of rounds.
- Advanced practical attacks (C, C++) on reduced round AES achieving >1000% speed-up through parallelization.
Microsoft
Software Engineer, Dynamics CRM | Hyderabad, India | July 2017 - September 2019
- Expanded and maintained Email Engagement feature in Azure Service Fabric for 1M+ active users.
- Advanced UI/UX for Sales Insights in ReactJS reducing average task completion time by 20%.
Education
Columbia University
- (2022 - 2023) Master of Science (Thesis Track) in CompSci
- GPA: 3.7/4.0
- Courses: Deep Learning, Natural Language Processing, Fundamentals of Speech Recognition, User Interface Design
Indian Institute of Technology Kharagpur
- (2012 - 2017) Bachelor of Technology & Master of Technology in CompSci
- GPA: 3.85/4.0
- Courses: Machine Learning, Artificial Intelligence, Foundations of Cryptography, Cryptography & Network Security
Projects
TODO: Update this more.
Cognate Detection - M.S Thesis
(2022-2023) Technologies: Transformers, Word2Vec, GloVe fastText, Kaldi, Python
- Created machine learning based cognate detection system for Avestan and Sanskrit word pairs, the first of its kind.
- Optimized the similarity metric computation using semantic, acoustic embeddings achieving >70% accuracy.
- Publication in progress
Automatic Speech Recognition (ASR)
(2022-2023) Technologies: Kaldi, Python, C++, Bash
- Optimized pronunciation approach for lexicon in ASR for Tamil and Telugu (in review for publication).
- Achieved significantly improved Word Error Rate for Telugu (16.65%) with respect to the baseline (19.45%).
Object Detection in Haze
(2022) Technologies: Computer Vision, YOLOv5, TensorFlow, PyTorch, Python
- Developed computer vision system for detecting vehicles in hazy images; Placed in Top 5 teams.
- Fine-tuned open-source image dehazing, object detection models achieving 70.1 mAP (mean Average Precision).
Stop Clickbait: Headline Classifier
(2015-2016) Technologies: Python, JavaScript, scikit-learn, NLTK
- Built machine learning based Chrome extension that detects and blocks clickbait headlines in social media.
- Scraped 20k+ headlines and trained Support Vector Machine (SVM) classifier models achieving 94% accuracy.
- Received Best Paper award at ASONAM 2016 Conference and national media coverage.
Chatterbox: Conversation Topic Classifier
(2014-2016) Technologies: Python, NodeJS, MongoDB, NLTK
- Designed and implemented classifier for text messages based on topic of conversation to make navigation easier.
- Developed prototype apps using MEAN stack and won multiple hackathons conducted by Microsoft, IBM.
Publications
View my Google Scholar profile for the most up-to-date list of publications.
Journal Articles & Preprints
- Ahaan Dabholkar, Sourya Kakarla, Dhiman Saha
- Arxiv Pre-print (In Review)
- Dhiman Saha, Sourya Kakarla, Dipanwita Roy Chowdhury
- IET Information Security (2019)
- Dhiman Saha, Sourya Kakarla, Srinath Mandava, Dipanwita Roy Chowdhury
- Journal of Hardware and Systems Security (HaSS 2017)
Conference Proceedings
- Sourya Kakarla, Srinath Mandava, Dhiman Saha, Dipanwita Roy Chowdhury
- International Conference on Applications and Techniques in Information Security (ATIS 2017)
- Dhiman Saha, Sourya Kakarla, Srinath Mandava, Dipanwita Roy Chowdhury
- Security, Privacy, and Applied Cryptography Engineering (SPACE 2016)
- Abhijnan Chakraborty, Bhargavi Paranjape, Sourya Kakarla, Niloy Ganguly
- Advances in Social Networks Analysis and Mining (ASONAM 2016)
- Best Paper Award
Awards and Recognition
- Andrew P. Kosoresow Memorial Award for Excellence in Teaching and Service, Columbia University (2023)
- Invention recognized by Columbia Tech Ventures: Employment Insights Dashboard (2022)
- Nominated for Prime Minister’s Research Fellows: Selected for final round (placed in top 6 CS students) (2020)
- Nominated for Best Master’s Thesis: Cryptanalysis of AES and AES based ciphers, IIT Kharagpur (2017)
- National media coverage: IIT team builds solution that weeds out ‘clickbait’ headlines (source) (2016)
- Best Paper, ASONAM 2016 Conference: Stop Clickbait paper (2016)
- Winner, IBM DAY 2016 - IIT Kharagpur exhibition/contest (2016)
- Winner, Microsoft’s Code.Fun.Do Finalist Forum nationwide hackathon, India (2015)
- Winner, Microsoft’s Code.Fun.Do hackathon, IIT Kharagpur chapter (2015)
- Runner-up, IBM’s Global Entrepreneurship Summit hackathon, IIT Kharagpur (2015)