Hello, I am
Ajay Therala
Building the future with Generative AI & Agentic AI Systems.
AI/ML Engineer with 4 years of experience & published researcher building production-grade RAG pipelines, Generative AI, and Agentic AI systems. Skilled in Machine Learning, NLP, LLM orchestration, scalable AWS deployments, and CI/CD Pipelines.
01. Education
Master of Science in Computer Science
Aug 2023 - May 2025Arizona State University, Tempe, AZ
GPA: 3.93 / 4.0
Coursework: Topics in NLP, Data Mining, Data Visualization, Statistical ML, Data Processing at Scale, Knowledge Representation & Reasoning
Bachelor of Technology in Information Technology
Aug 2017 - July 2021Jawaharlal Nehru Technological University, Hyderabad, TS, India
GPA: 9.21 / 10.0
Years Experience
IEEE Publications
02. Technical Arsenal
AI & Machine Learning
Cloud & Data Engineering
Programming & Development
Technology Stack
03. Experience
AI Engineer
Dexian LLC - Arizona State University (Client)
- Evaluated ChatGPT and Claude artifact-generation strategies, identifying patch-based updates as optimal; built a versioned artifact framework using Claude-style incremental patching, eliminating full-code regeneration during iterative web-app edits.
- Engineered dynamic conversation-history truncation, intelligently prioritizing 60% recent context and summarizing older turns to prevent token overflow across varied LLM context windows.
- Built a configurable web-crawling Lambda API that scraped sites N levels deep, parsed PDFs, and indexed content into OpenSearch while reducing redundant crawls through a 7-day DynamoDB cache.
- Orchestrated Canvas LMS syllabus synchronization with an SQS FIFO-driven Lambda, sequentially processing course events to eliminate duplicate sync failures and reliably re-index documents across 15,000+ courses.
- Reduced platform load time by 92% (2s to 150ms) by eliminating runtime DynamoDB lookups with in-memory JSON configurations for faster, lower-latency execution.
- Integrated a serverless LLM safety guardrail on AWS Lambda to detect prompt injection and jailbreak attempts across 10,000+ daily LLM transactions, using Meta PromptGuard-86M.
AI Engineer (Contract)
Predictive UX
- Engineered a production-grade AI skills extraction API leveraging prompt engineering & RAG for O*NET Top-K semantic mapping.
- Validated Skills Extraction API achieving 87%+ precision, 84%+ recall, & 92% skill observability, boosting verifiable credential reliability.
AI Full Stack Developer
Enterprise Technology at ASU
- Reduced AWS Lambda compute time and processing costs by 95% by replacing LangChain chunking with Semchunk, batching LLM embedding requests, and expanding document ingestion from PDF, DOCX, and TXT to 20+ file formats using Unstructured.
- Automated large-scale Canvas data onboarding by building an ingestion API that indexed 23,000+ course projects into Open Search.
- Engineered Jenkins-based CI/CD to build, test, Dockerize apps, publish images to ECR, and automate AWS Lambda deployments.
- Implemented cloud infrastructure using Terraform to provision and manage AWS Services.
Systems Engineer (ML Developer)
Tata Consultancy Services - TechBU, Digital Research & Innovation
- Engineered core components for a domain-specific search engine (Cognitive Product Support), enhanced search accuracy by 30%.
- Crafted Data Lens, a component for training custom NER models, achieving 80% - 90% accuracy.
- Developed and demonstrated GPT-powered conversational bots using prompt engineering, supporting over 5,000 daily user interactions.
- Led the development of an automated document digitization pipeline using AWS Textract for OCR processing, reducing manual efforts.
ML Research Project Intern
Tata Consultancy Services Limited
- Improved intrusion detection performance by 6% using data refinement and ensemble learning (Random Forest, XGBoost) techniques.
- Achieved a Top-5 global rank in the NetML-2020 challenge and presented research findings at ICAIIC 2022.
- Orchestrated end-to-end ML workflow in Django, showcasing strong machine learning engineering and deployment practices.
04. Projects
AI Resume Analyzer
AWS-powered analyzer using Bedrock LLM, API Gateway, and Lambda to assess job fit and skill gaps.
Yelp Dataset Analysis
Scalable ETL and data transformation using PySpark and Spark SQL on massive datasets.
LLM Hallucination Analysis
Benchmarked MLLMs across VQAv2/CLEVR datasets to identify robustness gaps in reasoning.
EEG Signal Analysis - Volunteer Research
Analyzed GBs of EEG signals, physiological and sensor data from 32 participants using statistical modeling & visualization.
05. Publications
What's Next?
Get In Touch
I'm passionate about building production-grade AI systems and always open to discussing new opportunities in AI/ML Engineering, Generative AI, and Agentic AI. Let's connect!