Amanbir Singh
Verified Expert in Engineering
Data Scientist and Back-end Developer
Amanbir has 10 years of experience in data science, analytics, and back-end engineering. He has worked at a large multilateral organization and with early-stage tech startups. Amanbir excels at working with clients in tackling complex business problems and has deep expertise in machine learning, data analysis, and building scalable web apps.
Portfolio
Experience
Availability
Preferred Environment
Python, Data Analytics, Data Science, Machine Learning, Pandas, Generative Pre-trained Transformers (GPT), OpenAI GPT-3 API, Minimum Viable Product (MVP), Generative Pre-trained Transformer 3 (GPT-3), OpenAI GPT-4 API, User Interface (UI), Product Management, Large Language Models (LLMs)
The most amazing...
...data science project I've worked on is building an automated machine learning platform for credit risk assessment from the ground up.
Work Experience
ML Developer
ATS Software
- Worked on a computer vision model to extract information from unstructured PDF files (including drawings, tables, etc.).
- Operated on NER models to extract information from natural language and unstructured text.
- Used GPT-4 for postprocessing of AI pipeline to improve the performance. Also included rule-based postprocessing to improve pipeline performance.
- Deployed the entire platform on AWS SageMaker and integrated with the client's stack.
- Trained multimodal models to improve NER performance.
Head of Product and Engineering
Monsoon CreditTech
- Led the development of the SaaS AutoML platform as an architect and product manager; made wireframes, wrote user and functional requirements, decided on back-end architecture, and ran sprints using Django, Angular, Jenkins, and Docker.
- Architected AutoML libraries used internally. The platform generated machine learning models optimized for lending.
- Acted as a product manager and architect for developer tools used by our internal data science team to speed up model development and deployment.
- Managed client engagements with 15 banks and NBFCs; built and deployed models to identify risky borrowers at the time of application. Increased revenue for the client by 20% and more.
- Hired and managed a team of 10+ data scientists and software developers. Conducted one on ones, set targets for the team, and mentored junior members.
- Built an auto-deployment process for machine learning models that supported multiple and multistage models.
Data Scientist | ML Expert
IISD Experimental Lakes Area Inc - Main
- Developed a model using meteorological data to predict the date of ice melt for a lake. The prediction was within a day of the actual ice melting date.
- Used boosting, bagging, and other algorithms to improve performance.
- Created a dashboard using React to show model predictions and performance.
Data Scientist
Independent Research Group
- Created a simulation to model the interactions between different economic actors (firms, employees, non-economic participants, etc.).
- Ran a Markov chain simulation to understand the effect with different initial states and interventions.
- Created output visualizations and statistics to test hypotheses.
AI/ML Developer
America Interpretation
- Developed a real-time translation API to convert speech to speech across any language.
- Built a back end in Django to handle streaming audio data and return translated audio data and transcription. The back end also addressed meeting creation and meeting joining.
- Created a front end in React and used RecordRTC to capture audio. Established a WebSocket connection to allow for audio streaming to the back end.
- Deployed both front and back end on Azure services.
- Integrated with multiple translation and speech generation services.
AI/ML Expert/Consultant
Harbor
- Did prompt engineering to improve LLM model predictions.
- Compared open-source LLMs against closed models.
- Self-hosted open-source LLMs on the company's infrastructure.
- Built a prompt testing framework in Python to compare and improve prompts.
AI/ML Engineer
Grown Unknown, LLC
- Developed prompts to generate customized parental advice using OpenAI APIs.
- Added context to the prompts to tailor the tone of the outputs.
- Compared OpenAI with other options and created a plan for future product development.
Machine Learning Expert
AmpVis Ltd.
- Advised the client on building the MVP, including all technical steps needed.
- Decided on team structures to handle different product decisions.
- Consulted on hiring decisions for other technical roles.
Data Scientist
NewCloud Medical LLC
- Built a Looker Studio dashboard to show data and summary statistics based on filters.
- Added visualizations in the Looker Studio to generate insights from the data.
- Created dashboard views that dynamically update based on selected fields.
Research Coordinator
JustJobs Network
- Set up an internal data management system to track versions of datasets.
- Led research on vocational training and skill-building programs in India. Led data collection and analysis; published a findings report.
- Designed a training module on statistics and R, which was used for the training of new hires.
Consultant
World Bank Group
- Supervised statewide data collection for 4,500 surveys at the individual and household levels.
- Built models to identify factors that affected education and labor market outcomes for adolescents.
- Participated in the dissemination of research findings.
Senior Research Associate
Centre for Microfinance Research
- Managed two randomized control trials studying the effect of financial access in India.
- Trained and supervised a field team of 30 members for 1,700 individual surveys across four districts.
- Designed and implemented six electronic questionnaires using Open Data Kit and SurveyCTO and built the back end for the survey data.
Experience
AutoML Platform for Lenders
https://monsoonfintech.com/thoth/The platform produced models for new applications and to help with collections for running loans. This was offered as a SaaS product.
Custom Machine Learning Models for Lenders
https://monsoonfintech.com/Built and delivered models to the largest lenders in India. This led to a 30% reduction in delinquencies and increased loan approvals by 25%.
Report for the World Bank
https://documents.worldbank.org/en/publication/documents-reports/documentdetail/866381523450216235/a-window-of-opportunity-a-diagnostic-of-adolescent-girls-and-young-women-s-socio-economic-empowerment-in-jharkhand-indiaMy role included experimental design, data collection, analysis, and modeling. I also worked on the dissemination of the report and communication with key stakeholders.
Skills
Languages
Python, HTML, R, SQL, TypeScript, CSS, JavaScript
Frameworks
Django, Django REST Framework, Bootstrap, Material UI, LlamaIndex, Angular, Flask, Scrapy
Libraries/APIs
Pandas, XGBoost, Scikit-learn, REST APIs, NumPy, Beautiful Soup, Sockets, Google Vision API, Amazon Rekognition, React, RecordRTC
Tools
Amazon SageMaker, ChatGPT, Git, Spreadsheets, Amazon Elastic Container Service (Amazon ECS), GitHub, Azure Machine Learning, Pytest, STATA, Open Data Kit, Azure ML Studio, Looker, Named-entity Recognition (NER)
Paradigms
Data Science, Automation, Object-relational Mapping (ORM), Object-oriented Programming (OOP), Agile, Microservices, Agile Software Development, ETL, Requirements Analysis, Unit Testing, DevOps, Agent-based Modeling
Platforms
Jupyter Notebook, AWS Lambda, Amazon EC2, Docker, Amazon Web Services (AWS), Azure, Azure Functions, Kubernetes, Google Cloud Platform (GCP)
Storage
MySQL, Amazon S3 (AWS S3), PostgreSQL, MongoDB, Databases, Database Architecture, Azure Cosmos DB, Azure Blobs, NoSQL, Amazon DynamoDB
Industry Expertise
Project Management, Banking & Finance
Other
Machine Learning, Data Analytics, Data Mining, Web Scraping, Artificial Intelligence (AI), Data Analysis, Statistics, Statistical Analysis, Predictive Analytics, APIs, Architecture, Automation Scripting, Scripting, Decision Trees, Data Scientist, Natural Language Processing (NLP), Regression, PDF Scraping, Scraping, Back-end, Software Architecture, Non-performing Loans (NPL), Data Scraping, Predictive Modeling, Customer Segmentation, Visualization, Full-stack Development, API Integration, Software Development, PyPDF2, Advisory, Technology Strategy & Architecture, Web Development, CTO, Technical Leadership, Generative Pre-trained Transformers (GPT), OpenAI GPT-3 API, Minimum Viable Product (MVP), Startups, Regular Expressions, Linear Regression, Data-driven Decision-making, Programming, Integration, Models, GPT, Exploratory Data Analysis, EDA, Modeling, Data Cleaning, Unstructured Data Analysis, Large Data Sets, Data Gathering, Machine Learning Automation, Data Processing, Back-end Development, Regression Modeling, Large Language Models (LLMs), OpenAI, Prompt Engineering, System Architecture, Product Roadmaps, Product Strategy, New Product Development, Team Leadership, Statistical Modeling, Unsupervised Learning, Supervised Learning, Machine Learning Operations (MLOps), Data Visualization, Data Reporting, Time Series, Time Series Analysis, Real-time Data, Leadership, Recommendation Systems, Serverless, AI Design, Full-stack, Solution Architecture, Data Structures, Generative Pre-trained Transformer 3 (GPT-3), Mathematics, Task Scheduling, OpenAI GPT-4 API, Decision Modeling, Neural Networks, Cloud, Language Models, Language Learning, Generative Systems, Product Management, LangChain, Speech to Text, Voice Recognition, FastAPI, Containerization, Retrieval-augmented Generation (RAG), Llama 2, Research, Cloud Computing, Unsupervised Fraud Detection, Generative AI, Open-source LLMs, Survey Design, SaaS, Optimization, Financial Modeling, Causal Inference, openpyxl, User Interface (UI), Deep Learning, AIOps, Graphics Processing Unit (GPU), Hugging Face, Pinecone, Text to Speech (TTS), Azure Text to Speech, Elementor, WebSockets, Vertex, Gradient Boosting, Google Earth, Markov Chain Monte Carlo (MCMC) Algorithms, Monte Carlo Simulations, Simulations, Computer Vision, Object Detection, Text Detection
Education
Bachelor's Degree in Economics and Statistics
Carnegie Mellon University - Pittsburgh, PA, USA
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring