Sai Ram Aditya | AI Engineer & Researcher

01 / about

Who I Am

I'm a Software Developer 2 at Oracle and AI researcher with a B.Tech in Artificial Intelligence from IIT Hyderabad. My work spans enterprise software, machine learning research, and production AI systems.

I specialize in Explainable AI, LLMs & Generative AI, and bridging cutting-edge research with real-world deployment. My first-authored paper on SHAP-based XAI has garnered 120+ citations.

When I'm not building LLM-powered enterprise workflows at Oracle, I'm exploring agentic AI, multi-agent orchestration, and cloud-native architectures.

120+ Citations

3 Publications

8.33 IIT GPA

2+ Yrs @ Oracle

profile.ts

// sai_ram_aditya.profile
const me = {
  role: "Software Developer 2",
  company: "Oracle",
  education: "B.Tech AI · IIT Hyderabad",
  expertise: [
    "Explainable AI",
    "LLMs / GenAI",
    "Agentic Systems",
    "Cloud Architecture",
  ],
  location: "Hyderabad, India",
  openToWork: true,
};

02 / experience

Where I've Worked

Software Developer 2

Oracle · Hyderabad, India

Jun 2023 — Present Current

Built a resume screening pipeline using LangChain and GPT-4o that parses candidate documents, extracts structured profiles, scores them against JD rubrics, and ranks candidates — reducing manual Tier-1 screening effort for HR teams.
Built an agentic HR email triage system using LangGraph that classifies inbound emails by category, urgency, and sentiment via a fine-tuned Mistral-7B model, auto-responds to FAQ queries through a RAG pipeline, and routes or escalates complex cases.
Integrated an LLM-powered conversational workflow enabling natural-language interaction with enterprise APIs through intent extraction and confirmation-gated operations.
Designed and exposed secure REST APIs and developed 10+ data-driven web UI pages integrating those APIs to support complex workflows and bulk operations.

LangChain LangGraph GPT-4o RAG Oracle ADF REST APIs Java

Data Science Intern

ABB GISPL · Hyderabad, India

Jan 2022 — Jun 2022

Modeled transformer power-output dynamics using SINDy and trained a Random Forest regressor to predict operating stress with SHAP-based feature attribution.
Trained LSTM time-series models on simulated operational data to forecast long-term transformer behavior across fleet-management scenarios.
Conducted research in explainable AI and applied machine learning, leading to multiple peer-reviewed publications.

SINDy Random Forest LSTM SHAP / XAI Python

03 / skills

Tech Stack

🧠AI / ML

TensorFlow PyTorch Scikit-learn LangChain LangGraph LlamaIndex ChromaDB OpenAI API HuggingFace XAI / SHAP LSTM Random Forest SINDy

⚡Languages

Python Java SQL JavaScript

☁️Cloud & Infra

GCP AWS Docker PostgreSQL

🔧Frameworks

FastAPI Streamlit Oracle ADF Oracle VB

04 / projects

What I've Built

🤖

Agentic RAG Research Assistant

Research assistant that autonomously fetches and answers over academic papers via a 3-strategy discovery pipeline (ArXiv title-match, ArXiv topic search, Semantic Scholar) with MMR diversity ranking, routed by a 6-field intent classifier.

Two-stage retrieval: text-embedding-3-small recall + BAAI/bge-reranker-large cross-encoder reranking 15→5 candidates
Event-driven hybrid memory combining a 5-turn rolling buffer with GPT-4o-mini prose summarization
Containerized with Docker + FastAPI backend

PythonLlamaIndexChromaDB OpenAIFastAPIDockerRAG

💬

LLM-Powered Enterprise Workflow

Conversational interface for Oracle enterprise APIs — natural-language interaction with business objects via intent extraction and confirmation-gated write operations.

NL-to-API intent extraction and classification
Safety-first confirmation-gated write operations
Integrated with Oracle ADF & Visual Builder

LLMsOracle ADF REST APIsJava

⚡

Transformer Stress Forecasting

Physics-informed ML system modeling transformer power-output dynamics using SINDy, with SHAP explainability and LSTM-based long-term stress forecasting across EV fleet scenarios.

SINDy for physics-informed equation discovery
SHAP-based model transparency and attribution
LSTM time-series fleet stress forecasting

PythonSINDy LSTMSHAP

05 / research

Publications

First Author

120+ citations

Enhancing trust and interpretability of complex machine learning models using local interpretable model agnostic SHAP explanations

International Journal of Data Science and Analytics · 2024

View on Google Scholar ↗

Co-Author

Efficient EV Fleet Management using Automated Thermal Stress Computation for Transformer in Electrical Distribution System

ICPEEV · 2023

View on Google Scholar ↗

Co-Author

An Integrative Method for COVID-19 Patients' Classification from Chest X-ray Using Deep Learning Network with Image Visibility Graph as Feature Extractor

CICBA · 2023

View on Google Scholar ↗

06 / credentials

Certifications & Education

☁️

Certification

GCP Associate Cloud Engineer

Google Cloud

Deploying, managing, and securing applications on Google Cloud with focus on automation, security, and AI/ML integration.

🤖

Certification

Oracle Generative AI Professional

Oracle

Specialized in LLMs, embeddings, prompt engineering, and responsible AI practices for enterprise applications.

🎓

Education

B.Tech, Artificial Intelligence

Indian Institute of Technology, Hyderabad

July 2019 — May 2023 · GPA: 8.33 / 10

🗾

Language

Japanese — JLPT N3

Japan Foundation

Passed July 2025 · Business Proficiency in English

07 / contact

Let's Connect

Open to new opportunities, collaborations, and interesting conversations about AI.

Email

sairamaditya88@gmail.com

→

GitHub

SaiRamAditya8

→

sairamaditya

→