Home

Tejareddy - AI/ML Engineer

[email protected]

Location: Fort Worth, Texas, USA

Relocation: yes

Visa:

Resume file: Tejareddy_AI_resume_1791377063578.docx Please check the file(s) for viruses. Files are checked manually and then made available for download.

TEJA REDDY Senior AI/ML Engineer | Generative AI & Agentic AI Email: [email protected] | Phone: +1 (940) 407-8601 | Tejareddy_Linkedin PROFESSIONAL SUMMARY AI/ML Engineer and Data Scientist with 11+ years across data engineering, machine learning, and enterprise AI, with recent specialization in Generative AI and agentic systems across finance, healthcare, and insurance domains. Designed multi-LLM and multi-agent architectures using LangGraph, LangChain, Azure OpenAI, and Azure AI Foundry, incorporating ReAct agents, Model Context Protocol (MCP), tool calling, intent-aware routing, specialist agents, targeted retrieval, response synthesis, and agent memory within a unified enterprise AI platform. Built RAG (Retrieval-Augmented Generation) and hybrid retrieval pipelines using Azure AI Search, combining vector embeddings, keyword and semantic search, query expansion, metadata filtering, ranking, and permission-aware retrieval to deliver grounded, context-aware responses with source citations. Engineered production-grade Python applications and backend microservices using FastAPI, Flask, PyTorch, TensorFlow, and Scikit-learn for enterprise AI/ML workflows. Applied Pandas, NumPy, Dask, Spark, and BigQuery to data wrangling, feature engineering, and large-scale distributed processing for analytics and model development. Integrated GPT-4o/GPT-5, Claude, Gemini, LLaMA 3, and Grok into conversational, text-generation, automated summarization, and insight-generation workflows through prompt engineering, using Hugging Face Transformers, SpaCy, and NLTK for supporting NLP and NER tasks. Developed multimodal AI solutions combining text, image, and speech processing with Azure Speech, Whisper, Stable Diffusion, and custom LLMs for cross-modal understanding and automation. Accelerated AI prototyping and proof-of-concept (POC) development using Builder.io, Databricks, PySpark, Streamlit, and Python to validate product concepts and technical feasibility, enabling rapid stakeholder demos and faster transition of POCs into production-ready solutions. Established internal responsible AI guidelines covering prompt design standards, governance, auditability, prompt-injection protection, and guardrail integration for enterprise LLM deployments. Deployed cloud-native AI/ML solutions across AWS (SageMaker, Lambda, ECS), Azure (Cognitive Services), and GCP (Vertex AI), using Terraform to provision infrastructure for scalable, secure workloads. Optimized data models, queries, and pipelines across PostgreSQL, MySQL, MongoDB, Redis, and Elasticsearch to support high-performance AI applications and enterprise data access. Implemented AI observability and evaluation workflows using Azure Monitor, Application Insights, OpenTelemetry, Prometheus, and Grafana, incorporating tracing, feedback analytics, and automated testing. Integrated AI models and Python microservices with Airflow, Spark, and Kafka pipelines to automate loan-processing and risk-management workflows, including real-time scoring. Implemented Agent-to-Agent (A2A) protocol-based communication to support autonomous collaboration, dynamic task delegation, and coordination among LLM-based agents. Operationalized MLOps practices through CI/CD with GitHub Actions, Jenkins, and GitLab CI, containerized deployments with Docker, Kubernetes, and Helm, and experiment tracking with MLflow and Weights & Biases. Delivered predictive analytics and real-time anomaly detection systems using XGBoost, Scikit-learn, and TensorFlow to identify fraudulent transactions, predict loan default risk, and reduce false positives, incorporating SHAP/LIME explainability and dashboards built with Tableau, Power BI, Matplotlib, Plotly, and Seaborn for performance and risk monitoring. Led cross-functional delivery and developer mentoring, translating business requirements into scalable AI solutions in collaboration with executives, data scientists, and engineering teams while incorporating GDPR and HIPAA requirements into deployment and operational practices. TECHNICAL SKILLS Programming Languages Python (Advanced), SQL, Scala, JavaScript (Basic), Shell Scripting AI / Machine Learning Frameworks TensorFlow, PyTorch, Keras, Scikit-learn, XGBoost, LightGBM, CatBoost, ONNX, Hugging Face Transformers Generative AI & LLM Ecosystem GPT-4/5, Claude, Grok, Gemini, LLaMA, Azure OpenAI, Azure AI Foundry, LangChain, LangGraph, ReAct Agents, Model Context Protocol (MCP), OpenAI API, Anthropic API, Guardrails AI, PromptLayer, LangSmith, LoRA/QLoRA, PEFT, AWS Bedrock Agentic AI & Orchestration LangGraph, ReAct Agents, Multi-Agent Orchestration, Tool Calling, MCP Servers/Tools, Skills, AI Agents, LLM Routing, Agent Memory, Scheduled AI Workflows, Human-in-the-Loop (HITL) RAG & Vector Databases AI Search, Pinecone, FAISS, Weaviate, ChromaDB, ElasticSearch, Vector Embeddings, Retrieval Evaluation (F1, Recall), RAGAS Data Science & Analytics Pandas, NumPy, SciPy, Dask, Spark MLlib, Databricks, Delta Lake, Feature Engineering, SHAP, LIME, Evidently AI (Drift Detection), Time-Series Forecasting (Prophet), Clustering, PCA Cloud & MLOps Platforms AWS (SageMaker, Bedrock, Lambda, ECS, S3, Redshift), Azure (OpenAI, AI Foundry, AI Search, Container Apps, ACR, Static Web Apps, Web PubSub, Document Intelligence, Blob Storage, Service Bus, Application Insights, Monitor), GCP (Vertex AI, BigQuery), MLflow, Weights & Biases, Kubeflow, CI/CD (GitHub Actions, Jenkins, GitLab CI), Docker, Kubernetes, Helm, Terraform Data Engineering & Pipelines Apache Spark, PySpark, Airflow, Kafka, Delta Lake, ETL/ELT Design, Data Validation, Feature Store (Feast) Visualization & Monitoring Tableau, Power BI, Plotly, Matplotlib, Seaborn, Azure Monitor, Application Insights, Grafana, Prometheus, Streamlit, Dash Databases Relational (PostgreSQL, MySQL) and Non-relational (MongoDB, Elasticsearch, Redis) Snowflake, Hive, Oracle 11g. Security & Privacy HIPAA, SOC 2, GDPR compliance in AI/ML workflows AI Infrastructure and Platforms NVIDIA CUDA (A100, T4), TensorRT, Mixed-Precision Training, Model Quantization, GPU Workload Profiling PROFESSIONAL EXPERIENCE Client: Accelerant, Sandy Springs, GA Aug 2025 - Present Role: Gen AI Engineer Responsibilities: Designed and developed Accelerant AI Hub, a centralized enterprise Generative AI platform and marketplace providing employees unified access to AI assistants, multiple LLMs, enterprise knowledge, AI agents, skills, connectors, MCP integrations, scheduled tasks, and internally hosted AI applications through a single platform. Developed the core agentic Retrieval-Augmented Generation (RAG) architecture using LangGraph, Azure OpenAI, and Azure AI Search, enabling grounded and context-aware responses across enterprise knowledge with source citations, session context, and conversational follow-up support. Implemented advanced hybrid retrieval combining vector embeddings, keyword search, semantic retrieval, metadata filtering, query expansion, ranking, and permission-aware search, improving the relevance and grounding of responses across distributed enterprise knowledge sources. Built LangGraph-based ReAct agent workflows with dynamic tool calling, enabling LLMs to solve user requests, invoke enterprise tools, perform iterative retrieval and generate context-aware responses for complex multi-step interactions. Developed a multi-LLM architecture using Azure OpenAI and Azure AI Foundry, providing configurable model selection that allows users to access and select preferred models including GPT, Claude, Grok through a unified AI Hub experience. Designed an advanced LLM orchestration and multi-agent architecture using LangGraph, enabling intent-aware routing across domain-specialist agents, MCP tools, connectors, and targeted RAG, with parallel agent execution, response synthesis, conversational context, and memory-assisted dispatch for complex enterprise queries. Implemented Model Context Protocol (MCP) capabilities within the AI Hub, enabling users to connect MCP servers, tools, and contextual resources and extend agent capabilities for enterprise and external-system workflows. Developed a reusable Skills framework supporting global and personal skills, AI-assisted skill generation and refinement, skill sharing, file-backed context, administrative promotion, and reusable execution patterns for specialized agentic AI workflows. Built an integrated Knowledge Base supporting document uploads, folder organization, sharing, indexing, re-indexing, and user/group-level permission-aware retrieval, enabling user-managed enterprise knowledge to be securely incorporated into RAG and agentic AI workflows. Developed the AI Hub Apps Marketplace, centralizing access to internally developed and externally hosted AI applications including Replit, Vercel, with capabilities for application discovery, categorization, favorites, usage tracking, and administrative approval workflows. Implemented enterprise-grade authentication and authorization using Microsoft Entra ID, OAuth 2.0, JWT validation, RBAC, and Microsoft Entra ID On-Behalf-Of flows for delegated Microsoft Graph access, enabling secure SSO and delegated access to organizational resources based on authenticated-user permissions. Designed permission-aware enterprise retrieval and security controls using Entra user/group identities, metadata-driven security trimming, session ownership validation, source-level authorization, query sanitization, prompt-injection mitigations, and access-aware retrieval to ensure users and AI agents access only authorized content. Developed scalable backend services using Python and FastAPI, integrated through MuleSoft as the enterprise API gateway layer, to securely expose and govern AI Hub services supporting agent orchestration, RAG, LLM interactions, streaming responses, enterprise integrations, Knowledge Base, skills, scheduled workflows, feedback, and platform administration. Contributed across the React-based frontend and Azure deployment architecture, implementing AI Hub user experience, model selection, agents, Knowledge Base, authentication, session history, file uploads, and streaming responses while deploying services using Docker, Azure Container Apps, Azure Container Registry, Azure Static Web Apps, Azure Web PubSub, and GitHub Actions. Contributed to the architecture of a hybrid local/cloud LLM inference strategy, designing policy-aware model routing based on data sensitivity, model quality, latency, capacity, and cost, with governed fallback between locally hosted open-weight models and approved cloud LLMs. Implemented production observability, evaluation, and AI governance using Azure Monitor, Application Insights, OpenTelemetry, structured logging, tracing, feedback analytics, and automated testing, while contributing to patterns for tool authorization, auditability and controlled agent actions. Environment: Python, FastAPI, React, LangGraph, LangChain, Azure OpenAI, Azure AI Foundry, Azure AI Search, GPT, Claude, Model Context Protocol (MCP), MuleSoft, Microsoft Graph API, Microsoft Entra ID, Azure Document Intelligence, Azure Blob Storage, Azure Service Bus, Azure Web PubSub, Azure Container Apps, Azure Container Registry, Azure Static Web Apps, Docker, GitHub Actions, Azure Monitor, Application Insights, OpenTelemetry, PostgreSQL, Redis, Replit, Vercel, REST APIs, Linux. Client: PennyMac, Westlake Village, CA Dec 2022 - July 2025 Role: AI/ML Engineer Responsibilities: Designed and deployed GPT-4 and LangChain-based workflow automation for loan origination and servicing, streamlining case processing and reducing manual review errors by 35%. Built RAG pipelines using Elasticsearch and Pinecone to retrieve relevant financial information and ground generated responses in loan-related source documents. Prototyped LLM-assisted semantic data discovery and contextual summarization within Databricks workflows, using Python microservices to demonstrate decision-support capabilities for loan and risk analytics. Developed and tuned financial risk and loan eligibility models using TensorFlow, PyTorch, and scikit-learn to support predictive analysis and risk assessment. Developed anomaly detection workflows to flag potentially fraudulent transactions for review and reduce false-positive alerts. Built FastAPI and Flask services to expose model predictions and automation capabilities, integrating Kafka for event-driven processing and Airflow for workflow orchestration. Applied SHAP and LIME to explain financial risk-model predictions, supporting model review and stakeholder understanding of factors influencing predictions. Implemented RAG evaluation workflows using precision/recall measures and BLEU/ROUGE reference comparisons to assess retrieval and generated-output performance. Integrated Guardrails AI for LLM response validation and PromptLayer for prompt experimentation and performance tracking. Automated model and service delivery using GitHub Actions and Jenkins, with Docker packaging and Kubernetes deployments; reduced model training time by 40% through GPU and data-pipeline optimization. Collaborated with Product, Data Engineering, Finance, Compliance, and Risk teams to develop Gen AI prototypes and production ML solutions aligned with business requirements and enterprise controls. Environment: Python, GPT-4, LangChain, TensorFlow, PyTorch, scikit-learn, Databricks, Elasticsearch, Pinecone, FastAPI, Flask, Kafka, Airflow, SHAP, LIME, Guardrails AI, PromptLayer, Docker, Kubernetes, GitHub Actions, Jenkins, NVIDIA A100 GPUs. Client: Elevance Health, Indianapolis, IN July 2020 - Nov 2022 Role: Machine Learning Engineer Responsibilities: Developed, trained, and tuned classification, regression, and clustering models using scikit-learn, TensorFlow, and PyTorch to support healthcare analytics and predictive modeling. Built deep learning models for medical image analysis and clinical text processing, using CNNs for imaging and sequence models for patient-record analysis. Led a cross-functional clinical document summarization prototype using transformer-based models, with Databricks and Azure Machine Learning supporting data preparation and model experimentation. Developed clinical NLP pipelines using spaCy, NLTK, and Hugging Face Transformers to extract named entities and structure information from unstructured medical records for downstream analysis. Automated training, validation, and deployment workflows using Airflow for orchestration and MLflow for experiment tracking and model management, supporting reproducibility and traceability. Containerized ML inference services with Docker and deployed models on Kubernetes and Amazon SageMaker to support healthcare application workloads. Implemented data and prediction drift monitoring with Evidently, alongside Prometheus and Grafana dashboards for inference-service latency and operational performance. Collaborated with clinicians, product managers, and DevOps teams to integrate ML outputs into healthcare platforms and support organizational HIPAA-related security and privacy requirements. Environment: Python, SQL, NumPy, Pandas, scikit-learn, TensorFlow, PyTorch, Hugging Face Transformers, spaCy, NLTK, Databricks, Azure Machine Learning, MLflow, Airflow, Docker, Kubernetes, Amazon SageMaker, Evidently, Prometheus, Grafana. Client: State of PA, Harrisburg, PA. Feb 2018 - June 2020 Role: Data Scientist Responsibilities: Developed and validated statistical models and machine learning prototypes using healthcare and public-sector datasets to support forecasting, trend analysis, and policy-related analytical decisions. Performed data profiling, cleansing, and feature engineering using PySpark, Pandas, and SQL, preparing datasets for model development, validation, and analytical reporting. Integrated Azure Machine Learning with Azure Databricks workflows to automate model retraining, validation, and deployment tasks for predictive analytics. Built ETL/ELT pipelines using Azure Data Factory, Azure Databricks, and PySpark to ingest and transform source data into datasets for machine learning and business intelligence. Implemented Delta Lake tables to maintain consistent analytical datasets supporting feature engineering, model training, and reporting workloads. Developed Spark Structured Streaming pipelines with Kafka to process incoming events and support incremental feature updates for downstream analytics. Supported migration of on-premises SQL Server and Oracle data into Azure-based database and analytical environments, preparing source data for reporting and machine learning workloads. Applied role-based access controls, managed identities, and Azure Key Vault to secure data workflows, with Azure Monitor and Log Analytics supporting operational monitoring and troubleshooting. Environment: Python, SQL, PySpark, Pandas, NumPy, scikit-learn, Azure Machine Learning, Azure Databricks, Azure Data Factory, Azure SQL, Azure SQL Data Warehouse (now Azure Synapse Analytics), Azure Data Lake Storage, Delta Lake, Apache Spark, Kafka, Azure Key Vault, Azure Monitor, Log Analytics, Power BI, Matplotlib, Seaborn, Git. Client: BNSF Railway | Fort Worth, Texas. Dec 2016 - Jan 2018 Role: Data Engineer Responsibilities: Engineered batch ETL pipelines using Python, PySpark, and SQL to transform railway operational data into datasets supporting performance analysis and recurring business reporting. Developed Kafka ingestion and Apache Spark streaming workflows to process operational events and make data available for near-real-time analytics. Built Amazon S3-based data lake workflows to store and process datasets in Parquet, Avro, and JSON formats for downstream analytics and reporting. Automated incremental data ingestion with validation checks, processing logs, and error-handling routines to support consistent and traceable pipeline execution. Collaborated with data scientists to prepare sensor datasets for predictive maintenance and anomaly detection, supporting Spark MLlib workflows and operational trend analysis. Implemented CI/CD workflows using Git and Jenkins and containerized data services with Docker, supporting deployment and management through Kubernetes to support scalable and secure production data pipelines. Applied AWS IAM policies, role-based permissions, and encryption controls to manage access to data storage and processing resources. Environment: Python, PySpark, Apache Spark, SQL, Hadoop, AWS (S3, Lambda, EMR, Redshift), Kafka, Docker, Kubernetes, Jenkins, Git, Tableau, PostgreSQL, Hive, Parquet, Avro, JSON, Linux. Client: Ford, India Aug 2014 - Aug 2016 Role: Data Analyst Responsibilities: Collected, cleaned, and transformed sales, operations, and marketing datasets using SQL and Python (Pandas, NumPy) to support recurring analysis and business reporting. Conducted exploratory data analysis to identify sales trends, demand patterns, and customer behavior, translating findings into insights for forecasting and inventory planning. Developed sales forecasting models using scikit-learn, analyzing historical sales data and demand patterns, preparing model inputs, and automated monthly forecast preparation to support inventory planning and business decision-making. Automated recurring reporting workflows using Python and SQL, reducing manual reporting time by 70%. Created Power BI and Excel dashboards, supported by Python visualizations, to communicate sales performance, demand trends, and operational findings to business stakeholders. Developed customer-feedback analysis prototypes using NLTK and spaCy, extracting sentiment trends and recurring themes to inform product discussions. Optimized SQL queries and Python scripts, documented analytical workflows in notebooks, and collaborated with business and engineering teams to incorporate analysis and forecast outputs into recurring reporting processes. Environment: Python (Pandas, NumPy, Scikit-learn, Matplotlib, Seaborn), SQL (MySQL, PostgreSQL), Power BI, Excel, NLP (SpaCy, NLTK), Jupyter Notebook, Git, Linux EDUCATION Bachelor of Computer Applications - IIMC, INDIA May 2014 Keywords: continuous integration continuous deployment artificial intelligence machine learning business intelligence sthree California Georgia Idaho Pennsylvania