Rahul Kumar
Verified Expert in Engineering
AI Engineer and Developer
Gurugram, Haryana, India
Toptal member since October 15, 2025
Rahul is a lead AI engineer with over seven years of experience building production-grade AI systems across Web3, fintech, HR tech, edtech, and identity intelligence startups. He has contributed over $150 million in startup growth, authored peer-reviewed research in IEEE and Springer, and holds patents in routing intelligence and predictive modeling. Known for transforming deep AI research into scalable, high-performance systems, Rahul delivers solutions that drive measurable business impact.
Portfolio
Experience
- Deep Learning - 7 years
- Python 3 - 7 years
- Machine Learning - 7 years
- Data Science - 7 years
- Large Language Models (LLMs) - 5 years
- Retrieval-augmented Generation (RAG) - 3 years
- Ray - 3 years
- AI Agents - 3 years
Preferred Environment
PyCharm, Slack, MacOS
The most amazing...
...solution I've built was an AI engine that automated more than 10 million background checks with 95% accuracy and delivered real-time insights.
Work Experience
Lead AI Engineer
Ferret.ai
- Developed Ferret AI's Identity Intelligence Engine, automating background checks across more than 10 million records with over 95% accuracy. Reduced manual review time by 70% and tripled API performance through a FastAPI migration.
- Built ETL pipelines processing over 100 million profiles into sharded MongoDB and Neo4j databases, enabling graph-based risk and association scoring. Built modular AI agents with multi-LLM orchestration, cutting research time from hours to minutes.
- Delivered interactive dashboards and dossier reports using Streamlit and a Notion-style interface, enabling investigators to access real-time risk scores and 360-degree profiles.
- Partnered with the UAE government, Morgan Stanley, and global enterprises to automate background verification, compliance, and fraud detection through Ferret AI's Identity Intelligence Engine.
Lead Machine Learning Engineer
O.XYZ
- Built the fastest routing intelligence, outperforming Meta LLaMA-3.1-70B, Qwen2.5-70B, and all open-source routers. Led routing research surpassing BBH, MMLU, MUSR, and GPQA benchmarks, powering Ocean AI with 20x faster performance than Perplexity.
- Deployed distributed LLM infrastructure on Ray Serve with H100 clusters, scaling hosted models for production workloads. Designed AutoEval LLM Judge and Jailbreak Guard to improve compliance, enhance security, and reduce hallucinations.
- Engineered optimized RAG pipelines delivering sub-second retrievals and knowledge grounding. Developed a marketing agent platform with on-chain block rewards and conducted foundation model research on Cerebras WSE and LiveKit Voice Agents.
- Published ORI research at the HK Summit and was featured in Forbes for breakthroughs powering Ocean AI.
Lead AI Engineer
Cognavi India Pvt
- Implemented RAG pipelines integrated with Neo4j and Groq LLM inference, enabling sub-second contextual retrieval and reasoning. Built FAISS-based retrieval for over 8.4 million job posts, cutting query latency by 65% and improving retrieval accuracy.
- Managed ETL pipelines processing 64+ million records with automated refresh cycles, reducing costs by 30% via SHA-based deduplication. Applied RLHF, PEFT, and LLM fine-tuning to boost model response quality by more than 20% on evaluation benchmarks.
- Built a data pipeline for 150+ million LinkedIn job posts and crafted AI-based job matching with patented digital profile technology, improving match accuracy by 35% and doubling engagement. Created a GPT-based resume builder and screening assistant.
Data Science Manager
Laytrip Inc
- Spearheaded data pipelines from AWS databases to BigQuery, aggregating over 300 million rows from multiple travel APIs on GCP. Designed predictive models for fare prediction, demand forecasting, and arbitrage optimization.
- Invented and patented an arbitrage model powering dynamic pricing. Implemented MLOps workflows for continuous training and deployment and automated Slack and Telegram bots, reducing manual operations by 40%.
- Created Cloud Function APIs for seamless partner integration. Collaborated with product, engineering, and operations teams, helping secure $300,000 seed funding from Airbus to scale the MVP and expand predictive analytics capabilities.
Founding Data Scientist
DataisGood
- Designed and developed ML, DL, and CV projects and courses, training more than 450 aspiring data scientists. Created quarterly roadmaps to expand course offerings, increasing learner engagement and platform adoption.
- Spearheaded career transition initiatives, affiliate programs, and university partnerships, expanding brand reach and credibility across global markets and driving measurable enrollment growth.
- Applied advanced analytics and KPI tracking to optimize business performance. The platform was later acquired by Skill Arbitrage for $3 million, validating the scalability and impact of these contributions.
Experience
Ocean AI Platform
Powered by Groq LPU acceleration and distributed Ray Serve clusters, Ocean AI delivers up to 20x faster inference, enabling real-time, multi-model orchestration across OpenAI, Anthropic, Mistral, and custom fine-tuned models.
Through continuous benchmarking on BBH, MMLU, and GPQA, ORI ensures every request is routed to the most capable engine—balancing speed, accuracy, and cost efficiency for optimal results.
Laytrip Predictive Booking
O Routing Intelligence
https://arxiv.org/abs/2502.10051Education
Postgraduate Degree in Data Science
California Institute of Technology - Pasadena, CA, USA
Bachelor's Degree in Computer Science
Lovely Professional University (LPU) - Phagwara, India
Skills
Libraries/APIs
XGBoost, OpenAI API, Claude API, REST APIs, WebRTC, PyTorch, TensorFlow, NumPy, Pandas, Pydantic, Asyncio, Python API, React, Hugging Face Transformers
Tools
Claude, ChatGPT, Amazon SageMaker, Amazon Elastic Container Service (ECS), AWS Glue, Amazon EKS, Looker, BigQuery, PyCharm, Slack, Prefect, Grafana, Gurobi, StatsModels, ARIMA, AutoML
Languages
Python 3, Python, SQL, JavaScript, TypeScript
Frameworks
LangGraph, AutoGen, Django, Agentic Frameworks, LlamaIndex, Ray, Next.js
Paradigms
Rapid Prototyping, Microservices, ETL, Testing, Model Context Protocol (MCP), Automation, Asynchronous Programming, Event-driven Architecture, Anomaly Detection
Platforms
Google Cloud Platform (GCP), Jupyter Notebook, LiveKit, CrewAI, Vertex AI, Amazon EC2, MacOS, Amazon Web Services (AWS), Docker, Azure, Databricks, Ollama, Kubernetes, LangSmith, Microsoft Power Automate
Storage
Data Pipelines, Amazon S3 (AWS S3), Neo4j, MongoDB, Elasticsearch, PostgreSQL
Industry Expertise
Healthcare, Banking & Finance, Cybersecurity, High-frequency Trading (HFT)
Other
Machine Learning, Deep Learning, Data Science, Large Language Models (LLMs), AI Voice Agents, LangChain, Google BigQuery, Machine Learning Operations (MLOps), Retrieval-augmented Generation (RAG), Artificial Intelligence (AI), Generative Artificial Intelligence (GenAI), OpenAI, Amazon Bedrock AgentCore, Data Analysis, Agentic AI, Cloud Services, Data Analytics, API Integration, Natural Language Processing (NLP), Web Scraping, Scraping, Data Collection, Text-to-Speech (TTS), Architecture, Technical Project Management, Startups, Technical Strategy, AI Tool Assessment, Cross-functional Collaboration, R&D, Solution Architecture, Tech Research & Evaluation, Custom Automation, Data Engineering, Sentiment Analysis, Prompt Engineering, Anthropic, Pattern Analysis, Image Generation, AI Chatbots, Conversational AI, AI Programming, AI Design, Full-stack, Workflow, Statistical Modeling, Attribution Modeling, Demand Forecasting, Marketing Analytics, Statistical Analysis, Predictive Analytics, Conversion, Funnel Marketing, Technical Leadership, Back-end Development, Software Architecture, Research, System Architecture, AI Automation, Data Handling, Data Modeling, Model Deployment, GPU Computing, Workflow Automation, Design, Reliability, AI Assistants, Agentic RAG Systems, ETL Pipelines, Forecasting, A/B Testing, Cloud Platforms, CSV, Unstructured Data Analysis, AI Agents, Hugging Face, Transformers, MLflow, FastAPI, Vector Data, Containers, Audio, Videos, APIs, Financial Systems, Compliance, Insurance Technology (Insurtech), Recommendation Systems, Statistics, Trading, Risk Management, Algorithmic Trading, Quantitative Analysis, Multimodal GenAI, Image Analysis, Markov Model, Multi-agent Systems, CTO, Product Delivery, Board Reporting, Finance, Fractional CTO, Amazon Kinesis, Scientific Data Analysis, Decision Support, Decision Support Systems (DSS), IT Management, RAG Pipelines, Vector Databases, Gemini, Model Validation, Spatial Analysis, Reinforcement Learning from Human Feedback (RLHF), Data Privacy, Identity & Access Management (IAM), Personally Identifiable Information (PII), Data Scientist, Marketing Mix Modeling, Bitcoin, Bayesian Inference & Modeling, Financial Engineering, Prediction Markets, AI Architecture, RAG Architecture, Webhooks, Model Evaluation, Tokenization, Embeddings from Language Models (ELMo), Time Series Forecasting, Multimodal Models, Supervised Fine-tuning (SFT), Fully Sharded Data Parallelism (FSDP), Vector Search, Agentic AI Systems, API Design, Performance Optimization, Performance, Microsoft Azure, Optical Character Recognition (OCR), Perplexity, Real-time Data Pipelines, Cloud Infrastructure, Document Parsing, HIPAA, Data Quality, Healthcare Software, Cursor AI, Industry.Trading, Data Analysis.Financial Data, Equity Trading, Large Language Model Operations (LLMOps), Open-source LLMs, Front-end Development, Quantitative Finance, Cryptocurrency
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