CV

Download résumé

Experience

• Aveni — Senior AI Engineer — Edinburgh, Aug 2024-present
  Building FinLLM - LLMs for the financial services industry.
  • Working across the whole pipeline – data filtering, pre-training, fine-tuning, pruning and distillation, domain adaptation, evaluation, retrieval-augmented generation – to build the FinLLM suite of language models.
  • Developing Agent Assure for automated safety and compliance assurance of AI agents in the financial services domain.
• Efficient Translation Limited — Deep Learning Engineer (part-time) — Edinburgh, Dec 2023-Apr 2024
  Corpus extraction and efficient low-resource machine translation.
  • Trained efficient machine translation and corpus cleaning models for low-resource language pairs.
  • Ran and optimised an efficient scalable parallel corpus extraction pipeline on web-scale data.
  • Delivered datasets and models to customers on time and meeting requirements

• University of Zurich — Visiting Researcher — Zurich, Mar 2023-May 2023
  Three-month visit, conducting research on detection and analysis of underspecification of the source sentence in machine translation, supervised by Dr. Rico Sennrich at the Department of Computational Linguistics.

• Amazon AWS AI — Applied Scientist Intern — Santa Clara, Nov 2022-Feb 2023
  Worked on isochronous machine translation for automatic dubbing. Co-organised the dubbing track at IWSLT 2023.

• TAUS — Data Engineer — Amsterdam, Jun-Oct 2020
  Worked on the EU-funded ParaCrawl project to collect parallel corpora from large-scale web crawls.
  • Optimised, maintained, and ran a highly scalable processing pipeline to extract, translate, align, and clean parallel corpora from web crawling data.
  • Consolidated and released the ParaCrawl corpus v7.0 and v7.1, comprising hundreds of millions of sentence pairs in many languages.
• Unbabel — Junior AI Researcher — Lisbon, Feb-Apr 2020
  Machine translation and quality estimation for customer-facing products.
  • Built domain-specific machine translation models.
  • Built quality estimation models to skip human post-editing for high-quality MT output.
• World Intellectual Property Organization (WIPO) — Fellow in Machine Translation — Geneva, Feb 2018-Jan 2020
  Development and maintenance of WIPO Translate and related NLP tools and technologies.
  • WIPO Translate: Built, improved, evaluated and deployed domain-specific neural and statistical machine translation models using the Marian and Moses toolkits.
  • IPCCAT: Developed neural text classification systems for patent categorisation.
  • Developed a system to retrieve semantically similar content from large collections of text using sentence embeddings and Faiss indexes.
  • Instrumental in the training and deployment of neural MT systems at several other international organisations and patent offices including IMF, OECD, WTO, IAEA, and KIPO.
• University of Edinburgh — Research Assistant — Edinburgh, Sep-Dec 2017
  Low-resource domain-specific machine translation research on the MeMaT project. Supervised by Kenneth Heafield and Alexandra Birch.
  • Worked on an EPSRC GCRF-funded project in collaboration with the University of Cape Town to build a machine translation system to facilitate communication in the medical domain between isiXhosa-speaking patients and English-speaking doctors in health centres in South Africa.
  • Collected corpora released as a public resource.

Education

• PhD in Informatics — University of Edinburgh — 2020-2024
  • 3-year doctoral programme at the Institute for Language, Cognition and Computation (ILCC), working with the StatMT group.
  • Working on using multi-encoder models to provide additional context to neural machine translation models to analyse and improve them. Download Thesis.
  • Supervised by Kenneth Heafield and Alexandra Birch.
• MSc in Informatics — University of Edinburgh — 2016-2017
  • 1-year postgraduate degree in Informatics, specialised in Natural Language Processing.
  • Graduated with distinction.
  • Courses included Machine Translation, Accelerated Natural Language Processing, Machine Learning and Pattern Recognition, Machine Learning Practical (Deep Learning).
• MSc in Computer Science — St. Xavier’s College, Kolkata — 2014-2016
  • 2-year postgraduate degree in Computer Science.
  • Graduated with 82% marks.
  • Courses included Artificial Intelligence, Data Mining and Warehousing, Image Processing and Pattern Recognition.
• BSc in Computer Science — St. Xavier’s College, Kolkata — 2011-2014
  • 3-year undergraduate degree with honours in Computer Science, and Mathematics and Physics as general subjects.
  • Graduated with 80% marks.

Also see: Projects

Technical Skills

• Python
• C++
• Git
• BASH
• Perl
• Docker
• AWS

Language Skills

• Bengali - Mother tongue
• English - Native
• French - Intermediate (B1)
• Hindi - Fluent
• Chinese (Mandarin) - Basic