
Rishab Pal
Verified Expert in Engineering
Data Scientist and AI Developer
Bengaluru, Karnataka, India
Toptal member since November 10, 2022
Rishab is an experienced data scientist and machine learning (ML) expert with a strong passion for mathematics and artificial intelligence (AI). He has 6+ years of experience designing and developing end-to-end machine learning solutions that have been successfully deployed to production on cloud platforms. Rishab constantly invests in his skills to stay up-to-date with the latest advancements in the field. He is committed to applying his knowledge and expertise to develop impactful solutions.
Portfolio
Experience
- Python - 7 years
- Natural Language Processing (NLP) - 7 years
- Data Science - 6 years
- Data Analytics - 6 years
- Statistical Modeling - 5 years
- Deep Learning - 5 years
- Software Deployment - 5 years
- DeepSeek - 1 year
Availability
Preferred Environment
Machine Learning, Deep Learning, Data Science, Natural Language Processing (NLP), Computer Vision, Python, Artificial Intelligence (AI), Generative Artificial Intelligence (GenAI), Machine Learning Operations (MLOps)
The most amazing...
...accomplishment I've had was showcasing my project at CES—International Consumer Electronics Show, the technology sector's flagship trade fair, in 2019.
Work Experience
AI Developer
Solstice Health Inc
- Extracted key medical claims from research documents and materials published by U.S. pharmaceutical companies, meeting FDA requirements.
- Used extracted data to create presentations for customers and clinicians, supporting sales pitches for new drugs.
- Built the solution on LangChain, with fine-tuned models and prompt engineering.
- Delivered over 90% successful claim extraction based on expert reviews and analysis.
AI Developer
FLIRT TECHNOLOGIES, LLC
- Developed AI-powered social media app features for personalized content creation using Stable Diffusion and Deepgram.
- Enabled near real-time immersive experiences with professional model photos, videos, and voice cloning.
- Used machine learning techniques like low-rank adaptation (LoRA) based on PEFT, model quantization, checkpoint merging, and other related approaches to achieve state-of-the-art (SOTA) optimization and performance.
AI/ML Expert
Nolea Technology Ltd
- Built an MVP for the recommendation system to match healthcare companies with clinical specialists in mental healthcare, allied healthcare, orthopedics, etc.
- Re-engineered the existing recommendation engine using SOTA NLP approaches and machine learning models like NER, LayoutLM, and LayoutParser, and built custom doc embeddings.
- Designed a cloud-native trainable and scalable architecture, previously built on a rule engine that lacked robustness for their business model.
LLM Chatbot Developer
Revibe AI
- Completed a POC of a mental health assistant built on top of Voiceflow. It understands the user's mind, responds with appropriate custom suggestions from the given knowledge base, or engages in a general conversation like a human.
- Built the RAG-based chatbot that replicates the behavior of a neuroscientist and emphatically engages with the user during the conversation.
- Used LangChain to enrich the vector store, adding proprietary neuroscience and psychological data to generate a more relevant response from the LLMs.
ML Expert | Data Scientist
Zachary Gidwitz
- Built a recommendation engine for an addiction recovery app that incorporates user demographics, mental state, and other metadata from app interaction to suggest quotes and prayers relevant to the user's state of mind.
- Fine-tuned the OpenAI GPT3 model and used Langchain to build a sequence of prompts where each preceding answer is included in the subsequent question. I added memory and linked the vector database quickly.
- Provided several offline and online metrics to track the live performance of the model and the recommendation system.
Machine Learning Engineer
Odem Global Pty Ltd
- Trained causal language model (CLM) with DeepSpeed on Bittensor, a peer-to-peer open-source protocol powering a globally distributed decentralized neural network.
- Used a mountain dataset (800+ GB) to train generative LLMs using DeepSpeed, which has lower network validation loss and thereby increases the model's reward, benefiting the client.
- Contributed to hyper-parameter tuning and revamped the default PyTorch data loader for faster dataset loading. Also, validated the models' performance on the validation set to ensure it meets the desired performance metrics.
- Fine-tuned models like GPT-J-Neo, Llama, MPT, Vicuna, Koala, Cerebras, to mention a few.
AI Expert
Rich Lemon Apps FZE LLC
- Developed a user sticker model using Stable Diffusion and ControlNet for a given style.
- Experimented with different configurations of parameters to understand the effect on the results and achieved state-of-the-art results.
- Explored different methods like checkpoint merging, trained a stickers model, and then fine-tuned that model on Portrait Face, and combined sticker style and user image training.
Senior ML Expert | Data Scientist
EagleView
- Revamped roof measurement automation for US houses by engineering a high-performance Line Detection model on aerial imagery to detect roof, wall, and ground lines from 28% to 59%.
- Developed segmentation models for roof and wall facet detection while handling obstruction, achieving 86% accuracy.
- Improvised complex PyTorch models to facilitate ONNX and TensorRT export for efficient deployment using NVIDIA Triton Server.
- Orchestrated microservices using Step Functions, AWS Lambda, and Amazon SageMaker endpoints, improving response time by 25%. This was followed by simplifying maintenance and support, focusing on maximizing code reuse.
Senior Software Engineer/Machine Learning (ML)
Airtel India
- Designed and built a chatbot, leveraging HuggingFace transformers with DeepSpeed to resolve queries for over 400 million of Airtel's monthly active users. Deployed successfully, handling over 10 million daily conversations with 85% accuracy.
- Improved the accuracy of face recognition, matching, and liveness-check models. Added face segmentation, blur and forgery detection, real-time blacklisting, and face search. Enhanced service efficiency and reduced manual effort by over 25%.
- Inculcated best MLOps practices on the team to lower technical debt and adapt workflow that requires little to no manual intervention for simulations, testing, deployment, and monitoring.
ML Engineer
Trantor Software
- Designed and deployed an OCR solution that outperformed the accuracy of Google OCR by 12% as benchmarked on the ICDAR 2013 dataset. Reduced the overall cost from an estimated $1 million to less than $50,000 per year.
- Prototyped a natural language processing (NLP)-based bot using recurrent neural networks to parse comments on the website, assess sentiments, and classify them into themes. Enabled standard responses and activation of the needed systems to respond.
- Worked on a trained convolutional neural networks model using INT8 fixed-point arithmetic. Designed algorithms to decide the Q. Collaborated to enable classifiers, detectors, generative adversarial networks (GANs), and segmentation networks.
- Trained a software categorization and brand recognition tool using word2vec, N-grams embeddings, and kNN classifier. This tool was capable of recognizing product titles and descriptions.
- Managed quantization-aware training of convolutional neural networks and worked on fine-tuning and post-training quantization.
Software Engineer
Yamaha Motor Solutions
- Conceptualized and built a ConvLSTM model on TensorFlow for hand gesture recognition. Used Bayesian optimization techniques for hyperparameter tuning.
- Worked on single-shot face recognition using SSD MobileNet for detection and Inception-ResNet v2 for feature extraction. Introduced TP-GAN for photorealism and identity, preserving the frontal face-view synthesis from any pose.
- Led a team to propose and implement predictive analysis for retail sales forecasting using ARIMA, exponential smoothing, and Holt's Winter models. This resulted in a 5.35% increase in quarterly sales.
- Implemented conditional Wasserstein generative adversarial network (cWGANs) to synthesize new vehicle designs from given samples.
Experience
Artificial Intelligence-based Document Capturing and Processing | AIDCAP
https://randomtrees.com/dcap2. Built and trained custom models for region proposal and text recognition with enhanced semantic understanding using Transformer.
Search Relevance and Ranking
https://www.airtelxstream.in/2. I alternated search keyword generation, increasing ad revenue by 20%. The system is scaled to handle >2,000 RPS using a Milvus-based vector search and is automatically refreshed daily. I then worked on a semantic search.
3. A custom deep semantic matching model trained for matching search queries with products. This resulted in a 10% growth in ad revenue. I also worked on search query classification using custom-trained embeddings for query understanding to match the search query with product leaf categories.
4. This project improved CTR by 50%, leading to a 37% incremental uplift in revenue and a 17% decrease in discount utilization. I improved the CTR using machine learning (ML)-based learn-to-rank models starting from simple linear regression and tree-based models, such as LightGBM and XGboost. These were combined with deep models for search-query understanding.
Callup AI | Chatbot Platform
https://www.airtel.in/contact-us2. It also comprised the nearest neighborhood-based intent detection solution using pre-tagged customer messages for tagging new customer messages to the support team.
3. The idea was to reduce customer service workload by automatically replying to user messages with pre-defined responses based on intent detection.
VirtualHR | Candidate Recommendation Engine
2. I automated the end-to-end process, leveraging the state-of-the-art (SOTA) NLP and machine learning approaches with cloud deployment.
Retail Sale Forecasting
https://www.yamaha-motor-india.com/The next step was incorporating Bayesian hyperparameter tuning with hybrid modeling using principal component analysis, support vector machines, and ensemble learning methods with Random Forests and XGBoost. This led to a 10% year-on-year increase in key metrics such as sales revenue.
Churn Prediction
The next step was ideating and implementing an A/B test framework and SQL queries. The framework defined key metrics to analyze model performance statistical significance.
Virtual Try-on
Face Dedupe Platform | Face Recognition, Matching, and Retrieval
https://www.airtel.in/bank/video-kyc2. I updated the platform with face segmentation, face blur detection, document forgery detection, real-time blacklisting, and face search to improve service efficiency and reduce manual effort by over 25%.
Price Optimization
https://tiki.vn/The work included data analysis, feature engineering, stakeholder management, designing A/B experiments, and performing statistical analysis for hypothesis testing.
Education
Bachelor's Degree in Computer Science
DIT University - Dehradun, India
Certifications
Deep Learning Specialization
Coursera
Machine Learning
Coursera
Big Data Modeling and Management Systems
Coursera
Programming the Internet of Things (IoT) Specialization
Coursera
Skills
Libraries/APIs
TensorFlow, PyTorch, Python API, OpenCV, Scikit-learn, Pandas, NumPy, SciPy, Natural Language Toolkit (NLTK), SpaCy, XGBoost, RAPIDS, REST APIs, OpenAI Assistants API, Google Vision API, Keras, Amazon Rekognition, DeepSpeed, OpenAI API, DeepSpeech
Tools
TensorBoard, Amazon SageMaker, Jupyter, Bitbucket, You Only Look Once (YOLO), OpenAI Gym, ChatGPT, Amazon Elastic Container Service (ECS), Dialogflow, NGINX, GIS, CPLEX, Apache Airflow, AutoCAD, Zapier, DeepSeek
Languages
Python, SQL, Python 3, Regex, Bash, Snowflake, JavaScript, Bash Script
Frameworks
Flask, Accelerate
Paradigms
Siamese Neural Networks, ETL, Microservices Architecture, Automation, Anomaly Detection, Microservices
Platforms
Linux, Jupyter Notebook, Docker, Azure, Amazon Web Services (AWS), Google Cloud Platform (GCP), Kubernetes, Mixpanel, Azure AI Studio, Databricks, Blockchain, Voiceflow
Industry Expertise
Project Management
Storage
Azure SQL Databases, PostgreSQL, MySQL, NoSQL, Google Cloud, Elasticsearch
Other
Machine Learning, Deep Learning, Natural Language Processing (NLP), Computer Vision, Software Deployment, Software Design, Statistics, Statistical Modeling, Data Analytics, Machine Learning Operations (MLOps), Image Processing, BERT, Long Short-term Memory (LSTM), Data Science, Predictive Modeling, Hugging Face, Artificial Intelligence (AI), Optical Character Recognition (OCR), Neural Networks, Recurrent Neural Networks (RNNs), Convolutional Neural Networks (CNNs), Artificial Neural Networks (ANN), Time Series Analysis, Entity Extraction, Classification, Image Recognition, Image Retrieval, Object Detection, Sequence Models, Machine Vision, Transformers, Transfer Learning, Graphics Processing Unit (GPU), Remote Team Leadership, Code Review, Generative Adversarial Networks (GANs), Clustering, Topic Modeling, Data Visualization, Mathematics, Optimization, Language Models, Recommendation Systems, Algorithms, Data Collection, Data Extraction, Data Manipulation, Large Data Sets, Analytics, Data Analysis, Models, Version Control Systems, Communication, Modeling, Data Engineering, Regression, APIs, Image Search, AI Design, Fine-tuning, Causal Inference, Data Scientist, Generative Pre-trained Transformers (GPT), Decision Trees, Chatbots, Cloud, OpenAI GPT-4 API, Chatbot Conversation Design, Integration, Programming, AI Programming, Data Cleaning, Unstructured Data Analysis, Machine Learning Automation, Language Learning, Image Analysis, Large Language Models (LLMs), Statistical Analysis, Data Reporting, Data Cleansing, API Integration, Prompt Engineering, Image Generation, Text to Image, Retrieval-augmented Generation (RAG), LangChain, FastAPI, Multi-agent Systems, Real-time Data, AI Agents, Distributed Systems, Multivariate Statistical Modeling, GPU Computing, Big Data, Web Scraping, Unsupervised Learning, Generative Pre-trained Transformer 3 (GPT-3), Revenue Optimization, Pricing Strategy, Handwriting Recognition, Stable Diffusion, NVIDIA TensorRT, Software Architecture, OpenAI GPT-3 API, Financial Modeling, Cognitive Behavioral Therapy (CBT), Data Structures, User Journeys, Customer Journeys, Amazon API Gateway, Generative Artificial Intelligence (GenAI), Point Cloud Data, Application Packaging, AI Model Optimization, Quantization, Internet of Things (IoT), Industrial Internet of Things (IIoT), Rankings, Computational Biology, Oncology & Cancer Treatment, Genomics, Molecular Biology, Biology, A/B Testing, Bittensor, 3D, Diffusion Models, ControlNet, Data Strategy, Data, User Interface (UI), Airtable, Web Applications, Natural Language Understanding (NLU), Point Clouds, Multimodal GenAI, Integrated Development Environments (IDE), Text to Image AI, DALL-E, OpenAI, Mistral AI, Llama 2, Llama 3, GraphRAG, Neural Network Pruning, Llama, FSDP, Parallelism, Amazon Bedrock, Claude, Gemini, NVIDIA TensorRT-LLM
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