Sam Patankar

Building a part-agentic AI pipeline to unlock the uptake of multifamily housing in Austin, Texas, from the Baseline workspace in Dublin. Admitted to the Baseline workspace on the strength of the work.

Building a part-agentic AI pipeline to unlock the uptake of multifamily housing in Austin, Texas. Admitted to the Baseline workspace, Dublin, on the strength of the work.

Worked on annotation strategy and dataset QA for computer-vision model training in the WeBuyBricks division. Improved automated line throughput by tightening data-feedback loops and continuously testing the pipeline.

WeBuyBricks is a new division of WeBuyBooks, which utilised automation hardware and software from the ground-up to generate value. I worked on annotation strategy and dataset QA for computer-vision model training; raised automated-line throughput ~300% through tighter data-feedback loops and continuous pipeline testing. (6-month contract — an extension of an initial 3-month engagement.)

Structured quality audits of model training data and outputs; rated model responses against RLHF/multitask benchmarks, flagging errors and edge cases to improve reliability.

Owned data validation and QA across a high-volume pipeline feeding an NLP text-analytics product; designed automated checks; coordinated, trained and mentored a graduate technical team; standardised processes across distributed teams.

First commercial technical role: Python, Scrapy, SQL/NoSQL, AWS, Flask; helped build the ETL platform feeding the NLP product. Rapidly promoted to Head of Data Operations.

Worked site-based in construction, progressing from apprentice to co-owner. Gained construction qualifications during this period.

Site-based construction and semi-skilled work; gained construction qualifications during this period.

Conducted wet-lab and clinical research to Good Laboratory Practice standards, including microscopy and imaging analysis of high-volume visual data. Co-authored a Nature Communications paper and a US patent application.

Wet-lab and clinical research to Good Laboratory Practice standards; microscopy and imaging analysis of high-volume visual data. Co-authored a Nature Communications paper and a US patent application.

Worked across financial services, telecoms, and customer operations, including large-scale data work sorting and cataloguing mortgage records. Also held customer-service and IT-support roles at mortgage-service firms, contact centres, and business customer services.

Following childhood years in the family retail and residential-property businesses, I spent my late teens and twenties across financial services, telecoms and customer operations. Beginning on a furniture production line (progressing to CNC machine operator), I moved to Bradford & Bingley Building Society, sorting and cataloguing over a century of mortgage records which was my first large-scale da

Semester 1 Modules:
- Speech Processing
- Accelerated Natural Language Processing
- Computer Programming for Speech & Language Processing
- Quantitative Methods for Linguistics & English Language
- [Audited] Applied Machine Learning

Semester 2 Modules:
- Natural Language Understanding, Generation, and Machine Translation
- Ethics of Artifical Intelligence
- Introduction to Bioinformatics
- Reinfo

Grade: Merit

Completed an MSc in Speech & Language Processing (Merit). Dissertation focused on tokenisation and sequence modelling, with hands-on work using PyTorch and HuggingFace.

Advanced Neuro- & Molecular Pharmacology