Headline

Senior Machine Learning Engineer @ Twilio | Published @ EMNLP | Ex: Level AI, Observe.AI, Neuron7.ai | Engineering Physics @ IIT Hyderabad ’22.

About

Engineering Physics Graduate from IIT Hyderabad with 4+ years of experience in Data Science and ML. My work at Observe.AI has spanned from Product-focused R&D to the end-to-end development and maintenance of API and queue-based microservices. This includes conducting load tests to determine service scaling policy, enabling robust monitoring and alerting, and ensuring service metrics (Latency, Throughput, Queue Wait Times, External LLM RPM/TPM, etc.) stay within desirable limits. Additionally, I've contributed significantly to several ML Platform reliability initiatives, enhancing observability and security of services. Our R&D efforts at Observe culminated in a paper published at EMNLP 2025, on the effective selective routing of queries to large LLMs, striking a balance between performance and cost. The published paper can be found here: https://aclanthology.org/2025.emnlp-industry.121/ At Neuron7.ai, I worked on all aspects of Search for Field Service Intelligence, personalizing and enhancing results grounded in customer Knowledge Base Data with domain-specific augmentation of queries and indexed data. Was also actively involved in optimizing the Data Ingestion Pipeline for both structured and unstructured data, and in enabling self-service for the Search PoC setup. Possess strong fundamental knowledge of Machine Learning Algorithms and Deep Learning architectures, and specialise in tackling NLP challenges across Supervised and Unsupervised, Predictive and Generative use cases. I believe in choosing the right tool for the right task with the end users' experience in mind. Why finetune an LLM with Billions of parameters when TFIDF can do the job? Experience collaborating with cross-functional teams, both tech and non-tech, on features, client projects, and deployments.

Location

Bengaluru

AI stack

claude code

cursor

langchain

openai

githubcopilot

huggingface

Work experience

Senior Machine Learning Engineer at Twilio, May 2026 - Present · 6 mos

Senior Machine Learning Engineer at Level AI, Jan 2026 - May 2026 · 5 mos, Noida, • Led the 0->1 architecture design, development, and productionization of Policy AutoQA, enabling contact centers to move towards automated QA grounded in company policies, beyond traditional conversation-based QA. • Performed distributed multi-task finetuning and benchmarking (on H100s & GB300) of Dense (Gemma-4B, Gemma-12B) and MoE (gpt-oss-20b) LLMs via GPT-5 distillation with DeepSpeed Zero-3 optimization for contextual query expansion and QA verdict generation. • Owned end-to-end service development and deployment with load testing to optimize service throughput (+75%) and latency (p99: -50%) using queue-based pipeline parallelism and online FP8-quantized KV Cache vLLM model serving. • Conducted extensive RnD for optimized embedder and reranker model serving using TEI, Infinity, Ray-serve and vLLM to enable cross-request batching and minimize GPU contention. • Developed a self-serve Gradio app for offline service eval and continual performance improvement.

Machine Learning Engineer at Observe.AI, Aug 2024 - Dec 2025 · 1 yr 5 mos, Bangalore Urban, ∗ Led exhaustive experiments for LLM-based Question Answering with rubric-driven evidence generation and reasoning - carefully curating data from Snowflake SQL DBs & AWS S3 to address extreme class skew - achieving +18% Macro-F1 and +15% Accuracy over in-house baseline using an SFT-finetuned Phi-4 model, while also benchmarking against external LLMs (GPT-4o-mini, Nova-lite). ∗ Productionized Nova-lite LLM optionality with in-house model, successfully onboarding multiple strategic accounts with robust alerting and extensive load tests for service scale, TPM & RPM. Currently serving ~90K requests/min and ~100M tokens/min. ∗ Developed a PoC for agentic AutoQA with CrewAI, integrating call transcripts, metadata, and screen recordings as custom tools to answer questions; achieved 68% Accuracy on in-house annotated dataset. ∗ Developed and deployed an end-to-end queue-based microservice for Machine Translation to enhance platform multilingual capabilities, conducted load tests to determine HPA scaling policy and set up robust monitoring & alerting on New Relic. ∗ Fixed a critical AWS Lambda security vulnerability in HIPAA-compliant cluster, encrypting credentials exposed in environment variables into the config. Additionally, merged the manual onboarding and automatic refresh Lambdas to streamline and unify data ingestion with reduced Infra cost and overhead. ∗ Enhanced ML microservices observability across the stack by fixing service availability computation in New Relic, integrating ELB 5xx errors alongside Target 5xx to accurately reflect service downtime. ∗ Migrated plaintext Cryptography-encrypted credentials from service configs to AWS Secrets Manager for enhanced security and exposure risk mitigation. Affected by org-wide restructuring.

Data Scientist, Search at Neuron7.ai, May 2022 - Aug 2024 · 2 yrs 4 mos, Remote, • Automated Incremental Data Ingestion for Search with a single centralized Self-service microservice that seamlessly handles concurrent requests from multiple clients. This helped significantly reduce PoC turnaround times and developer involvement. • Reduced time taken for table extraction from PDFs by 85% using Multiprocessing in Python. • Enhanced Solr retrieval for Search by augmenting user queries and indexed records with hierarchical or non-hierarchical metadata extracted from raw client data and enabling metadata-based search. Automated metadata extraction and tagging of indexed records using Cron jobs. • Leveraged LLMs for RAG-based generation grounded in client Knowledge-Base data and for summarization of complex/verbose user queries. • Generalized and integrated client-specific Structured Data Ingestion and Search modules as part of mainstream pipelines with suitable feature flags and configurable parameters. • Leveraged Google’s Document AI for OCR on scanned PDFs, extracting tables and text with near-perfect accuracy. Also performed Form Field extraction to answer objective queries from PDFs. • Developed a modular Python Pipeline with unit tests to perform grouped Agglomerative Clustering using Scikit-Learn with finetuned Sentence Transformer embeddings for deduplication of predictor-specific observation and resolution texts by directly fetching data from Solr cores. Improved the average cluster cosine similarity by 60% and significantly reduced the runtime from many hours to a few minutes. • Deployed Flask and FastAPI endpoints as services on Azure and GCP Linux environments. • Collaborated effectively with Cross-functional teams for Client Project Deliveries, Bug Resolutions, and Production Deployments. Led end-to-end Search implementation for 2 major clients, while contributing to multiple other successful PoCs.

Data Science Intern at Akaike Technologies, Nov 2021 - Apr 2022 · 6 mos, Remote, 1. Developed an Open Domain Question Answering System using Haystack Framework, based on Dense Retrieval. Uses the FAISS Document Store and Dual Encoder for queries and passages. Pretrained transformer models from HuggingFace were used for the encoders. Integrated a regular Closed Domain QnA system as well in the same project. 2. Developed an automatic answer evaluation system for factual and descriptive answers. Finetuned a DistilBert model on TREC dataset for classifying questions into above categories. The former uses NER and subject-object dependency for evaluation, while the latter employs a combination of Sentence-Bert cosine similarity and BLEU score between the reference and candidate answers. Both above projects developed on PyCharm IDE; also created Streamlit demos as PoC. 3. Performed classification of various room images into 6 distinct categories by finetuning a MobileNet-V2 model on the ADE20k dataset (sourced from Kaggle). Prepared detailed documentation for all projects.

Education

B.Tech, Engineering Physics, Indian institute of technology Hyderabad

Intermediate, MPC, FIITJEE

Public links

github: https://github.com/raj26000

leetcode: https://leetcode.com/raj26000