Generative AI & Cybersecurity AI Engineer

JIYA SIWACH

AI Engineer · Generative AI · LLMs · RAG · Agentic AI · Machine Learning · AI for Cybersecurity Delhi, India · +91-782******* · **************@*****.*** · linkedin.com/in/jiya-siwach-9b5826310 · github.com/jiyasiwach CORE SKILLS

Generative AI & LLMs

RAG · Hybrid retrieval (vector +

BM25, RRF) · Prompt engineering ·

LLM evaluation (Hit@k, MRR,

LLM-as-judge) · Guardrails &

prompt-injection defense · AI agents /

agentic workflows · Tool calling ·

LangChain · LangGraph · Hugging

Face · OpenAI API · Embeddings

Machine Learning & NLP

Scikit-learn · PyTorch · TensorFlow ·

Text classification · TF-IDF · Feature

engineering · Cross-validation ·

Anomaly detection

Data & Vector Stores

Python · SQL · Pandas · NumPy ·

Matplotlib · ChromaDB · FAISS ·

MongoDB · MySQL · Oracle

Backend & Deployment

FastAPI · REST / SSE · Node.js ·

Express · React · TypeScript ·

Docker · Git/GitHub · CI/CD · Render

· Vercel · AWS (basics) · OCI

Security AI

MITRE ATT&CK · OWASP Top 10 ·

CISA KEV · Log analysis · Threat

intelligence · Digital forensics · VAPT

basics

CERTIFICATIONS

• Generative AI Landscape — Infosys

Springboard, 2025

• Introduction to Cybersecurity —

Cisco Networking Academy, 2026

• OCI Application Integration

Professional — Oracle, 2025

• Database Management Systems —

Infosys Springboard, 2025

• Network Topologies & Technologies

— Skillsoft, 2025

• Product & Brand Management —

NPTEL, 2025

LEADERSHIP

• PR & Media Head, MRISA – Manav

Rachna InfoSec Army (2024 –

present)

• General Secretary, HackWithIndia

Student Chapter

• Student Ambassador, SET CSE

Specialization, Manav Rachna (2025

– present)

• Core Organizer, ENIGMA

Cybersecurity CTF & Learn IT Tour

2025

LANGUAGES

English (fluent) · Hindi (native) ·

German (beginner)

SUMMARY

AI Engineer building production Generative AI and ML systems at Magppie Wellness Kitchen — LLM/RAG pipelines, agentic workflows, voice AI, FastAPI services, evaluation harnesses and cloud deployment, shipped end to end. B.Tech in Computer Science (Cyber Security & Digital Forensics) with a focus on grounded, guardrailed and measurable AI for security.

EXPERIENCE

AI Engineer Jun 2026 – Present

Magppie Wellness Kitchen, Delhi NCR

• Built an AI automated call system combining agentic workflows, LLM integration and voice AI to run customer and sales calls, cutting manual calling effort for the sales team.

• Designed RAG knowledge flows with prompt engineering, guardrails and structured outputs so the agent answers product queries accurately from company data.

• Lead architecture and development of the corporate Learning & Development portal that centralizes and automates employee training.

• Developed generative-AI applications and predictive ML pipelines on industrial business data to optimize operational logistics and support data-driven decisions. Data Science & Analyst Intern Jul 2025 – Jun 2026

ShadowFox

• Built predictive analytics models on engineering datasets with Python, Pandas and Scikit-learn; earned Exceptional Performance honors for production-aligned ML pipelines and EDA.

Intern – Forensic & Cyber Laws Jun 2026

Delhi Police

• Trained on cybersecurity frameworks and digital-forensics workflows; analyzed data integrity, digital-evidence protocols and validation reporting. Full Stack Developer Intern Jun 2025 – Jul 2025

TechnoHacks Online

• Built a React.js dashboard on the OpenWeather REST API with optimized real-time state rendering.

PROJECTS

CyberSec RAG Copilot — grounded security Q&A with citations Oct 2026 cybersec-rag-copilot-tan.vercel.app · github.com/jiyasiwach/cybersec-rag-copilot · Python, FastAPI, ChromaDB, BM25, LLM via OpenRouter, React, TypeScript, Docker, Render, Vercel

• Built and deployed a RAG assistant that answers only from MITRE ATT&CK (706 techniques), OWASP Top 10 and CISA KEV — 948 indexed chunks — citing the exact source passage for every claim and refusing instead of hallucinating.

• Implemented hybrid retrieval (ChromaDB vectors + BM25, Reciprocal Rank Fusion, exact-ID boost): 100% Hit@5, 1.0 MRR on a 20-question set vs 0.87 MRR for BM25 alone.

• Designed a two-layer prompt-injection guardrail (weighted regex heuristics LLM classifier), retrieved context delimited as data: 5/5 attacks blocked, 0 false positives.

• Streaming answers over SSE with per-answer latency, token and cost metrics (~3 s,

~$0.001/query); LLM-as-judge eval harness and dashboard (faithfulness 1.0, citation accuracy 1.0); Dockerized with rate limiting and CORS. AI-Powered Log Anomaly Detection System

Python, Machine Learning, Cybersecurity AI

• Real-time log analysis and anomaly detection that parses system logs, detects behavioral deviations and flags security threats with visual alerts. EDUCATION

B.Tech, Computer Science & Engineering (Cyber Security & Digital Forensics) 2023 – 2027

Manav Rachna International Institute of Research and Studies (MRIIRS), Faridabad

Contact this candidate