Open to Internship & Full-Time

Ishita
Acharya

Final-year CS engineer building AI systems that ship to real users from clinical ML pipelines to voice AI to full-stack SaaS.

7.92/10
GPA - MITE, Karnataka
6+
Projects
AI Engineer Full Stack MERN LangChain RAG
Arsenal

What I build with

Full-stack from model to deployment, Python and ML on one end, React and cloud infra on the other.

🧠
AI / ML
PyTorchTensorFlowScikit-learn SMOTERandom ForestIsolation Forest OpenCVMediaPipeTesseract OCR Anomaly Detection
🌸
LLM & RAG
LangChainHugging FaceRAG OllamaChromaDBSentence Transformers BM25Hybrid SearchPrompt Engineering OpenAI APIMulti-Agent Systems
🎙️
Voice & Conversational AI
Speech RecognitionElevenLabs TTS / pyttsx3Conversational AI Spotify API
Backend
PythonNode.jsFlask FastAPIREST APIsJWT Auth LaravelRBAC
🌿
Frontend
React.jsJavaScript HTML5CSS3Bootstrap Streamlit
☁️
Cloud, DevOps & Data
AWS EC2S3MongoDB Atlas MySQLGit / GitHubRailway VercelUbuntu / LinuxJest GitHub Actions
🔐
Security
MITRE ATT&CKLog Analysis Intrusion DetectionThreat Triage Vulnerability Assessment
Work

Things I've shipped

From clinical AI to production SaaS — built independently or in small teams.

AI / RAG Apr 2026 – Present

Multi-PDF AI System

A privacy-focused RAG application for intelligent Q&A across multiple PDFs with source citations — fully offline, no cloud APIs required.

  • Hybrid retrieval: BM25 keyword search + ChromaDB vector search + Reciprocal Rank Fusion
  • Streamlit UI + CLI supporting PDF ingestion, summarisation, table extraction
  • Local LLM inference via Ollama fully offline document intelligence
PythonLangChain ChromaDBOllama BM25Streamlit
Security AI Jun 2025 - Jun 2026

AI Security Log & Incident Agent

Multi-layer log analysis pipeline that detects anomalies and auto-classifies incidents against MITRE ATT&CK in plain English.

  • Regex parsing + Isolation Forest anomaly detection 68% structuring coverage on mixed-format logs
  • LangChain + Hugging Face LLM agent for MITRE classification and severity scoring
  • Graceful degradation to rule-based fallback on LLM failure
PythonLangChain Hugging FaceIsolation Forest MITRE ATT&CK
Research ML Mar – Nov 2025

Peripheral Arterial Disease Detection

ML pipeline reading raw clinical Doppler reports and automatically flagging PAD, replacing manual clinician review.

  • Tesseract OCR → feature extraction from unstructured clinical PDFs
  • Random Forest + SMOTE - 75% accuracy, strong minority-class recall
  • Hybrid engine combining ML output with ABI clinical rule logic to cut false negatives
PythonScikit-learn SMOTEFlaskTesseract OCR
Personal Nov 2025 – Jan 2026

Voice-Controlled Music Player

End-to-end voice AI with real-time Spotify control zero touch required.

  • SpeechRecognition + ElevenLabs for command capture and natural voice feedback
  • Spotify API - play, pause, skip, volume control via voice only
  • Modular Flask backend built for low-latency, extensible pipelines
PythonFlask ElevenLabsSpotify APISTT
Personal Feb 2025

Zoe - AI Voice Assistant

JARVIS-inspired assistant with context-aware conversation, real-time voice I/O, and task execution.

  • OpenAI API for contextual, multi-turn conversational responses
  • ElevenLabs + pyttsx3 for high-quality TTS with local fallback
  • Flask web app wrapper for real-time command execution and integrations
PythonFlask OpenAIElevenLabspyttsx3
Credentials

Certifications

Coursework and industry simulations beyond the degree.

🏦
Cyber Job Simulation
Deloitte
🔐
Cybersecurity in Product Dev Lifecycle
MITE
🛡️
Vulnerability Assessment & Exploitation
MITE
⚔️
Prompt Wars
MITE

Contact

Let's build
something.

Open to AI/ML roles, full-stack positions, and anything where shipping actually matters.