Debanjan Mahata — AI Researcher

I'm a Senior Machine Learning Research Engineer specializing in building intelligent assistants using agentic AI—systems that can autonomously reason, plan, and act to support complex user workflows. My current focus is on developing agentic memory systems that enable long-term contextual reasoning and continuity in enterprise-grade applications. I am also very interested in making vision retrievers work for document AI workflows at enterprise scale and building working memory for Agentic AI applications that can manage the context seen by an user assistant Agent in diverse scenarios. Previously at Bloomberg, I developed cutting-edge Document Intelligence systems powered by large language models (LLMs), Retrieval-Augmented Generation (RAG), and multimodal AI, enabling financial data extraction and analysis at scale for global clients. I hold a Ph.D. in Integrated Computing, with a research focus in natural language processing, information retrieval, and scalable AI infrastructure. My work has been published at premier AI conferences, and I remain actively involved in the research community primarily as an author. Day to day, I collaborate with cross-functional teams—including software engineers, researchers, product managers, and ML engineers—to propose, prototype, and deploy impactful AI solutions. I also mentor Ph.D. interns and drive in-house research initiatives aimed at shaping the future of applied machine learning. In order to know more about Machine Learning at Bloomberg and the type of problems that we are solving and implementing please refer: https://www.techatbloomberg.com/post-topic/data-science/ Google Scholar: https://scholar.google.com/citations?user=8F1SwO0AAAAJ Homepage: https://sites.google.com/a/ualr.edu/debanjan-mahata/ (I don't maintain this anymore. Will come up with a new website soon.)

Stackforce AI infers this person is a Fintech-focused AI specialist with expertise in document intelligence and natural language processing.

Location: New York, New York, United States

Experience: 15 yrs 3 mos

Skills

Retrieval-augmented Generation (rag)
Large Language Models (llm)
Software Development
Deep Learning
Natural Language Processing (nlp)

Career Highlights

Expert in agentic AI and intelligent assistants.
Proven track record in document intelligence systems.
Published researcher in natural language processing.

Work Experience

Bloomberg LP

Senior Machine Learning Research Scientist (3 yrs 8 mos)

Moody's Analytics

Director of AI (1 yr 1 mo)

Bloomberg LP

Research Scientist (3 yrs 8 mos)

Infosys

Senior Research Associate (2 yrs 2 mos)

MyCityWay

Data Scientist Intern (2 mos)

Center of Innovation and Commercialization

Graduate Assistant (2 yrs 7 mos)

Graduate Teaching Assistant (4 mos)

Graduate Research Assistant (1 yr)

Cognizant Technology Solutions

Programmer Analyst Trainee (9 mos)

Education

Doctor of Philosophy (PhD) at University of Arkansas at Little Rock

BS at Banaras Hindu University

High School at Birla High School

Debanjan Mahata

AI Researcher

New York, New York, United States15 yrs 3 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Expert in agentic AI and intelligent assistants.
Proven track record in document intelligence systems.
Published researcher in natural language processing.

Stackforce AI infers this person is a Fintech-focused AI specialist with expertise in document intelligence and natural language processing.

Contact

Skills

Core Skills

Retrieval-augmented Generation (rag)Large Language Models (llm)Software DevelopmentDeep LearningNatural Language Processing (nlp)

Other Skills

Alchemy APIAlgorithmsApache KafkaApache PigApache SparkArtificial IntelligenceBig DataCCOBOLComputer ScienceComputer VisionCross-functional Team LeadershipDB2/SQLData GovernanceData Mining

About

Experience

15 yrs 3 mos

Total Experience

2 yrs 6 mos

Average Tenure

3 yrs 8 mos

Current Experience

Bloomberg lp

Senior Machine Learning Research Scientist

Oct 2022 – Present · 3 yrs 8 mos · New York, United States · On-site

Spearheaded the development of advanced retrieval-augmented generation models, enhancing document understanding capabilities.
Architected and implemented scalable end-to-end AI solutions, significantly improving information extraction processes from financial documents.
Collaborated with cross-functional teams to integrate large language models and multimodal approaches, driving innovation in Document AI.
Technologies and Skills that I explored:
Agentic AI, Retrieval Augmented Generation, Document AI, Information Extraction, Large Language Models, Multimodal Models for Document Understanding, LLM Engineering (Prompting, Finetuning - Instruction, SFT, RLHF, DPO), Architecting and Implementing robust end-to-end AI solutions that scales.

Retrieval-Augmented Generation (RAG)Deep LearningSoftware DevelopmentMicroservicesCross-functional Team LeadershipComputer Vision+11

Moody's analytics

Director of AI

Aug 2021 – Sep 2022 · 1 yr 1 mo · New York, New York, United States

Deep LearningSoftware DevelopmentExtract, Transform, Load (ETL)MicroservicesCross-functional Team LeadershipProject Management+8

Bloomberg lp

Research Scientist

Nov 2017 – Jul 2021 · 3 yrs 8 mos · Greater New York City Area

I worked as a Research Scientist at Bloomberg. My role at Bloomberg allowed me to work at the intersection of natural language processing (NLP), information retrieval, machine learning and software engineering. I was not only responsible for researching some of the challenging problems related to these areas, but also to build real-world solutions around them. The resulting applications shipped as products in the Bloomberg Terminal, enabling our clients around the globe to make smarter, more informed decisions about their business and financial strategies.
In order to know more about Machine Learning at Bloomberg and the type of problems that I might be involved in solving and implementing please refer: https://www.techatbloomberg.com/post-topic/data-science/

Deep LearningSoftware DevelopmentExtract, Transform, Load (ETL)Apache SparkMicroservicesPyTorch+6

Infosys

Senior Research Associate

Aug 2015 – Oct 2017 · 2 yrs 2 mos · Palo Alto, California

Worked in the R&D team of Infosys Information Platform (http://www.infosys.com/information-platform/)
Work Highlights
1. Developed a patented system for Automatic Ranked Keyword Extraction from Text Documents
2. Developed an efficient technique for producing phrase vectors using neural language modeling
3. Developed effective techniques for recommendation and search using keyword embeddings
4. Improved techniques of query expansion, indexing, content-based recommendation for building high quality search and information retrieval applications for text
5. Developed Microservices for NLP and Text Mining to be used internally by the team and the organization.
6. Developed Python backend for NLP and Text Mining capabilities on top of Spacy.
Things that I dealt with Daily
Text Clustering, Keyword Extraction and Ranking, Text Search, Recommendation and Query Expansion, Word Embeddings, Deep Learning applied to NLP (CNN, RNN), REST APIs
Technologies that I used:
Keras, Spark, Solr, Python, Java, MongoDB, Spacy, Docker
Research Areas:
Deep Learning, Machine Learning, Information Retrieval, Text Mining, Natural Language Processing

Deep LearningMicroservicesPyTorchPandas (Software)Natural Language Processing (NLP)

Mycityway

Data Scientist Intern

May 2013 – Jul 2013 · 2 mos · West Village, New York

Working as a part of the core team for developing MyCityWays’ next generation analytics and learning platform.
Finding new and innovative ways to improve customer segmentation, targeting and engagement.
Optimizing data quality, search quality, and predictive capabilities.
Natural Language Processing, Machine Learning Algorithms, Data Mining.
Contextual Ad targeting, Relevant content delivery.
Python, NLTK, Neo4j, MySQL.
Worked at the very initial stages of the MobileROI platform: http://www.mobileroi.com/index.html

Deep LearningNatural Language Processing (NLP)

Center of innovation and commercialization

3 roles

Graduate Assistant

Jan 2013 – Aug 2015 · 2 yrs 7 mos

Helping faculty, students and staff to protect and realize the full commercial potential of their innovations.
Assessing newly invented technologies, startup ideas and finding possible markets for them.
Discussing the possible opportunities of commercializing an invention.

Deep LearningNatural Language Processing (NLP)

Graduate Teaching Assistant

Aug 2012 – Dec 2012 · 4 mos

• Graded and taught HTML, CSS, Java Script to an undergraduate class of 32 students.

Natural Language Processing (NLP)

Graduate Research Assistant

Aug 2011 – Aug 2012 · 1 yr

Developed crawlers for crawling social media data and used programming APIs of different social media websites for collecting data.
Performed social network analysis, data mining and sentiment analysis on the collected data.

Natural Language Processing (NLP)

Cognizant technology solutions

Programmer Analyst Trainee

Jun 2010 – Mar 2011 · 9 mos · Kolkata, India

Developed Java and COBOL programs to read XML files generated by web-based applications and storing them to DB2 database.
Developed Java and COBOL programs for implementing new policy changes in life insurance applications of Metlife, USA.