Mohammed Aqid Khatkhatay
ML/AI engineer with 4+ years of experience spanning applied modeling, experimentation, and production ML systems. Built recommendation, NLP, conversational AI, and GenAI solutions serving 4M+ users, including hybrid traditional-ML + LLM systems using RAG, prompt engineering, and SFT; hands-on with Azure Databricks/AKS, real-time and batch inference, CI/CD, APIs, streaming, and observability.
Experience
Senior Machine Learning Engineer
ServiceNow Inc.
New York, NY
- Lead AI personalization and recommendation systems for Service Now University serving 4M+ users, improving recommendation CTR by 10%, increasing course enrollments, and raising CSAT by 5%; build hybrid recommenders combining matrix factorization/user similarity with GenAI using RAG, prompt engineering, and SFT.
- Architect real-time and batch inference pipelines where Adobe Analytics events trigger Kafka-based recommendation pre-computations; serve predictions through FastAPI with Redis and Azure Cosmos DB.
- Deploy models and inference workloads to Azure Databricks and AKS; package services with Docker/Helm and automate build, test, and release using Jenkins and Cloud Bees CI/CD.
- Expose ML APIs through Azure Kong Gateway using OpenAPI and implement Lightstep/Splunk monitoring and alerting for latency, failures, reliability, and incident response.
Machine Learning Engineer
ServiceNow Inc.
New York, NY
- Designed, trained, evaluated, and productionized traditional ML and Generative AI solutions for customer-facing applications, including Azure Databricks deployment and containerized inference on AKS.
- Built Rasa/OpenAI/open-source LLM assistants and semantic-search/intent-recognition workflows; applied RAG and prompt engineering to connect natural-language requests with enterprise data and actions.
- Implemented RAG-based Genius AI Search to retrieve courses, certifications, skills, cases, and knowledge articles from natural-language user queries, improving search relevance and discovery across Service Now University content.
- Developed FastAPI inference services, Jenkins/Cloud Bees CI/CD, Kubernetes Helm deployments, Kong/OpenAPI integrations, and continuous production monitoring with Lightstep.
- Integrated Python applications with Snowflake and automated schema discovery/metadata synchronization, reducing manual maintenance and improving AI workflow consistency.
Data Scientist
IBM
New York, NY
- Built enterprise conversational-AI/ML applications in Python and integrated watsonx capabilities into ML and GenAI workflows.
- Conducted model-fairness research with theoretical and empirical guarantees; developed tooling to distill qualitative model outputs into human-interpretable insights for chatbot migration.
- Applied text clustering and PCA to consolidate 120K+ intents/examples and evaluated anomaly-detection models for IVR logs, improving call containment by 10%.
Data Scientist Research Intern
Audible (Amazon)
Newark, NJ
- Worked with the Data Science Content Team to identify high-potential movie-adaptation candidates across a 6M+ book catalog; formalized AWS batch ETL pipelines spanning Kindle, Goodreads, and Audible data.
- Automated ingestion and processing of 2+ TB of Kindle, Goodreads, and Audible data using Spark and AWS Step Functions, ECR, Batch, Lambda, and Cloud Watch; aligned 60K books and movies using NLP, distributed Spark, and Redshift.
- Reduced 1,000+ features with UMAP, LDA, PCA, and t-SNE on Sage Maker; parallelized workloads on a 96-core EC2 instance and packaged and deployed scripts, pipelines, and models on AWS.
Education
Columbia University
B.E. Computer Engineering
GPA 3.7/4.00 CGPA 9.7/10, Top 1%
Skills
PythonPyTorchTensorFlowscikit-learnNLPrecommendation systemsanomaly detectionclusteringPCAfeature engineeringsemi-supervised learningmodel evaluationfairnessLLMsRAGprompt engineeringsupervised fine-tuning (SFT)embeddingssemantic searchconversational AIOpenAI APIsRasawatsonxAzureAKSKubernetesDockerHelmJenkins