PORTFOLIO

SARIKA SHIROLKAR

2025

TABLE OF

ABOUT
ME

Building real-world AI systems — from research prototypes to deployed automation. I build practical AI and machine learning systems that move from idea to real deployment. My work combines intelligent automation, data engineering, and cloud-ready applications. I enjoy turning complex problems into structured solutions through experimentation and clean design. IEEE-published researcher. Hackathon winner. Driven by curiosity and research, I focus on creating AI tools that are reliable, useful, and built to last.

Sarika's Assistant — a RAG Chatbot

Ask a question or choose a prompt.

Oct 2025 — Present

AI Engineer

AI Workflow Automate

Built and shipped Linkyro, a solo-built AI-powered Chrome extension (live on Chrome Web Store) that generates context-aware comments using LLMs.
Built and deployed agentic AI systems — voice scheduling agents using Retell AI + LLM orchestration with n8n.
Owned AI product development end-to-end: problem framing, AI pipeline design, model integration, production deployment.
Designed AI evaluation pipelines including error analysis, failure-case reduction, and performance benchmarking.

Mar 2025 — Sep 2025

Software Engineer (Cloud & AI Infrastructure)

AI Workflow Automate

Designed and deployed production backend services on Azure Linux VMs with focus on scalability and failure recovery.
Improved development velocity by ~40% through AI-assisted workflows while maintaining code review standards.
Standardized deployment practices (environment config, logging/monitoring, rollback-friendly releases).

Jul 2025 — Sep 2025

AI & ML Intern

Bharat Electronics Limited

Developed production-grade computer vision systems for defense platforms — real-time object detection with robustness under adverse conditions.
Optimized deep learning models (YOLOv8) on custom-annotated datasets for high-accuracy, low-latency inference.

Apr 2025 — Sep 2025

Project Intern

IEEE IAMPro'25

Developed an object identification and classification system for crime scene imagery as part of IEEE IAMPro'25.

PROFESSIONAL
EXPERIENCE

Selected Projects

PROJECTS

Linkyro — AI Comment Generator

Solo-built, live AI-powered Chrome extension that generates context-aware comments using LLMs.

#Chrome Extension#LLMs#Live Product

Kisan Mitra — Dual-Purpose AI Assistant

Chat-first AI product for farming families: crop advisory + scholarship discovery. Built with Databricks.

#Databricks#Spark SQL#LLM

Ebbinghaus Adaptive Memory Agent

Spaced-repetition app with AI-driven adaptive scheduling. FastAPI + SQLite + Streamlit + n8n.

#FastAPI#n8n#Streamlit

appointo.ai

End-to-end appointment platform for service businesses with booking, rescheduling, and cancellation.

#Next.js#AI Agents#Scheduling

IBM Sales Risk Prediction

ML pipeline for identifying high-risk sales outcomes. Preprocessing, feature selection, model training.

#Python#ML#Risk Analytics

Amazon Business Risk Prediction

Business risk prediction focused on transactional and operational signals.

#Python#Classification#BI

Retell Calendar MVP

Voice-agent MVP integrating Retell with calendar operations for automated scheduling.

#Retell AI#n8n#Google Calendar

DAWNVision

Computer vision for object identification under practical constraints.

#CV#YOLO#Deep Learning

Crater Detection

Detection pipeline for identifying lunar and Martian craters from image datasets.

#OpenCV#Image Processing

AI Research Agent

Agentic research assistant that gathers, summarizes, and structures outputs.

#LangChain#Streamlit#LLM

Portfolio Website

This portfolio with responsive layout, animated interactions, and deployment.

#HTML#CSS#JavaScript

Research

PUBLISHED
PAPER

IEEE IAMPro'25

Secure Object Identification Techniques for Autonomous Vehicle

First Author — IEEE Conference Publication

Secure object identification is a cornerstone of safety-critical perception systems in Autonomous Vehicles, particularly when navigating unstructured environments. This paper proposes a detection framework leveraging YOLOv8 to enhance visibility and classification reliability under adverse weather conditions such as fog, rain, haze, and low light. The model demonstrates a superior trade-off between inference speed and detection accuracy compared to traditional multi-stage detectors.

Skills

TECHNICAL STACK

Built for production-focused AI work: practical modeling, strong deployment basics, and reliable data workflows.

Languages

PythonSQLJavaC

AI/ML

LLMsTransformersClassificationFeature EngineeringModel EvaluationError AnalysisFine-tuning

GenAI / LLM Tooling

LangChainn8n OrchestrationPrompt EngineeringRAGTool-use Agents

Deep Learning & CV

CNNsYOLOv8TransformersOpenCVObject Detection

Libraries/Frameworks

scikit-learnTensorFlowKerasPandasNumPyFastAPIStreamlit

Cloud & Deployment

Microsoft AzureVMsDockerDatabricksLinux

Data & Visualization

Power BITableauAnalyticsPreprocessing

Databases

MySQLMongoDBSQLiteSpark SQL

Hackathon Wins

2nd Place — Databricks Hackathon. 1st Place — HackMarch 2.0. Builder mentality validated in competitive, time-constrained environments.

IEEE First-Author Publication

Peer-reviewed paper on applied object detection for autonomous systems at an IEEE International Conference.

Chair, IEEE CIS SVIT

Led ML workshops, hackathons, and mentored peers on applied ML pipelines and experimentation.

Academic Excellence

CGPA 9.1 — Ranked 2nd in department (Sem 6, 2024-2025).

Achievements & Leadership

RECOGNITION

Contact

LET'S BUILD
SOMETHING
USEFUL

Open to software engineering roles and collaborative AI product work.

sarikashirolkar@gmail.com linkedin.com/in/sarikashirolkar

Resume