Building a generative AI startup for real time conversational AI agents over telephone.

Founder

• Processing cohort of lung images with COVID with Super-Resolution GANs.

• Continued research into deep learning methods on OCT scans.

• Attempted commercialization of research in the detection of COPD in lung scans.

Research Assistant

Worked for UKs third largest insurance company.

• Designed and built the entire back-end system into an event-sourced microservice architecture to serve brand with over 4 million customers.

• Developed event-sourcing microservices architecture using Python and Redis, as well as the TDD, BDD, and Domain Driven Design paradigms.

• Aided infrastructure development on AWS, utilising services such as API gateway, Elasticsearch, and Fargate.

• Adopted the Agile methodology along with CICD for automated deployment.

Senior Software Engineer

AssetVault is a Fintech startup looking to safeguard your physical and digital assets throughdigital Wills and simpler insurance claims. 

Role was to develop multiple products, pitch C level members of largest insurance companies and investors, as well as manage end-to-end implementation of products by ensuring team of 7 devs and designers operated successfully.

Head of Engineering

Selected for Europe’s best accelerator (Investor AllStars 2016) specialising in deep tech.

EF is backed by Reid Hoffman (founder of LinkedIn) and Greylock Partners, Mosaic Ventures, Founders Fund, Lakestar and some of the world's best entrepreneurs.

Co-founded and developed multiple hardware and software business ideas.

Actively involved in researching markets, developing ideas and executing on both product and customer development to build end-to-end products.

Pitched and won contract with Peabody Trust, one of the UKs largest housing association, to use apartment/renter matching product developed by my start-up.

Entrepreneur in Residence

Have been using this everyday from building websites to scraping large search engines to developing novel machine learning prediction systems for cardiac OCT images.

As part of a research team at Imperial College London (top 10 unis worldwide) and Columbia University, published research in ISBI (top bio-medical journal) on using transfer learning to segment a proprietary dataset of cardiac OCT images, achieving clinician level performance and results.

Additionally, developed an end-to-end system to enable renters to find properties that better suited them using data from a variety of sources including the largest housing websites in the UK.

Currently continuing research in diagnosing COVID from Lung Scans in partnership with Imperial University.


Studied AI methods including propositional logic and abductive methods including HAIL and TAL. 

NOTE: ML is a subset of AI, this represents symbolic AI.

Managed team of 7 developers and designers to produce end-to-end products used by largest insurance company in UK.

Designed and built insurance platform (BE and infra) from idea conception through to delivery, used by 4M users daily.

Written and utilised many Web APIs including Google, Facebook, UK Government APIs, OpenStreetMap, Zoopla etc.

Developed R&D Alexa chatbots for insurance products in partnership with largest insurance company in UK.

Secured cloud servers for banking level encryption at rest and in transit on AWS. 

Created and implemented BCM (backup, contingency and monitoring) plans for data storage from large scale enterprise customers.

Launched multiple servers for deep learning computations on Google Cloud.

Wrote multiple bash scripts to ensure setup was automated and succesful.

Written multiple bash scripts to automate multiple routines including web scraping, server setup and installations.

Using Git for all code management for over 6 years in both private and professional life.

VC Backed Startup Entrepreneur (YC/Techstars) with AI research from Imperial College London (Top 10 Uni Worldwide)

Sim Virdi

Implemented FC-CNN to perform the segmentation of tissue layers from a dataset of cardiac OCT images using transfer learning.

Reviewed and ranked various CNN options to pursue task.

Adapted and augmented dataset to perform labelling task.

Wrote Bash scripts to automate server setup across Google App Engine resulting in having 10 separate machines, each with 2 NVIDIA Tesla K80 GPUs, operating at the same time for free.

Matched human level performance by achieving average dice score of 82% across all layers.

Built in Python Notebook using PyTorch, Tensorflow and Keras.

 
Segmentation of Tissue Layers on OCT images via CNNs

Developed an API to give accurate pricing valuations of any asset in the UK.

Programmed in Python using BeautifulSoup and Selenium.

6 month timeline completed in 2 weeks.

Universal Valuation API

Extracted, merged and cleaned financial, zip and gender datasets.

Applied exploratory data analysis (EDA) techniques to explore dataset for distributions, outliers and anomalies.

Performed Univariate, Bivariate and Multivariate analysis using various summaries and plots.

Programmed in R with results reported using R Markdown.

 Exploration of NY State Financial Contributions for the 2016 Presidential Elections

Extracted OpenStreetMap data for the City of London.

Performed data munging techniques, such as assessing the quality of the data for validity, accuracy, completeness, consistency and uniformity, to clean the data using Python.

Analysed data in MongoDB using PyMongo.

Reported results using IPython Notebook.

 Wrangle OpenStreetMap Data - London

Developed a BCI headset capable of recognising brain activity, enabling communication and control of an external
device.

Designed signal processing circuitry and fabricated PCB.

Classified the EEG output using standard pattern classifiers and data analysis techniques using MATLAB & Python.

 Brain Computer Interface

• Worked on a greenfield project, developing an event-sourced microservice architecture serving all aspects of the sale, support, and logistics for the entire brand and its 4 million customers.

• Programmed in Python using Redis, BDD paradigm, and Flask.

• Infrastructure from AWS including API Gateway, Elasticsearch, and Fargate.

Personal details

Sim V.

Summary

Technical skills

Work Experience

Personal Projects