class YashikaSoni:
def __init__(self):
self.name = "Yashika Soni"
self.pronouns = "She/Her"
self.location = "Jaipur, Rajasthan, India 🇮🇳"
self.education = "BCA (IT) @ Biyani Girls College, Jaipur (2023–2026)"
self.role = "AI/ML Engineer | Continuous Learner"
self.focus = ["NLP", "LLMs", "Data-Driven Solutions"]
self.currently = "Learning RAG Pipelines & Fine-tuning LLMs"
def say_hi(self):
print("Let's build real-world AI together 🚀")🤖 Jarvis Chatbot Assistant — Iron Man HUD-Powered AI Desktop Assistant
A voice-enabled AI desktop assistant featuring a full Iron Man HUD interface — animated rotating rings, live system diagnostics, a cinematic boot sequence, and Claude AI powering every conversation. Not just a chatbot — a personal command center built with Python.
| Attribute | Details |
|---|---|
| Stack | Python 3.10+, Anthropic SDK (Claude Sonnet), tkinter, customtkinter, pyttsx3, SpeechRecognition, psutil, python-dotenv |
| Scale | Desktop application · 20-message conversation memory window · Fully offline TTS |
| Performance | Instant built-in command resolution (no API) · Sub-second voice transcription via Google Speech API · 1.5s live system stats refresh |
| Security | API key isolation via python-dotenv · Key never touches source code · .env.example pattern for safe onboarding |
| Impact | Demonstrates modular AI architecture · Separates concerns across voice, commands, and LLM layers · Production-grade conversation history management |
| Repository |  |
Jarvis implements a smart routing layer: built-in commands resolve instantly without any API call, while unrecognized input routes to Claude Sonnet with full conversation context preserved. The Iron Man HUD delivers a cinematic UX with real-time CPU, RAM, and battery telemetry, animated ring states (standby → thinking → listening), and a timestamped log interface — all built natively in Python without Electron or a browser runtime.
📅 AI Timetable Generator — Constraint-Aware Academic Scheduling System
An intelligent timetable generation system that automatically allocates classes, rooms, teachers, and time slots while enforcing academic and logistical constraints. Designed for schools, colleges, training centers, and any structured learning program at scale.
| Attribute | Details |
|---|---|
| Stack | Python, FastAPI (fastMain.py), HTML5, CSS3, Vanilla JavaScript, JSON/CSV data layer |
| Scale | Multi-role system (Admin / Faculty / Student) · Multi-section / multi-batch / multi-semester support · Flexible time slot granularity (30 / 45 / 60 min) |
| Performance | Constraint-aware scheduling engine · Hard vs soft constraint separation · Room capacity + resource matching (Lab vs Lecture Hall) |
| Security | Role-based access control across Admin, Faculty, and Student portals · Separate login flows per user type |
| Impact | Eliminates manual scheduling conflicts · Supports pluggable solvers (ILP / Heuristic / Evolutionary) · CSV/JSON export for downstream integration |
| Repository |  |
The system is structured around a three-layer architecture: input parsers handle courses, teachers, rooms, and constraints; a solver core applies optimization strategies; and exporters produce human-friendly outputs. The frontend is role-segmented with dedicated dashboards, and the Python backend exposes a FastAPI service layer for on-demand schedule generation. Pluggable solving strategies include Integer Linear Programming, Backtracking with Pruning, Genetic Algorithms, and Simulated Annealing.
| Recognition | Details |
|---|---|
| 🥇 IBM Certified | Artificial Intelligence & Advanced IT Skills — Apr 2026 |
| ⚡ MUJ HackX 3.0 | 36-Hour International Hackathon @ Manipal University Jaipur — Nov 2025 |
| 🤖 AI Chatbot — Biyani College | Built Python + Rasa-powered assistant for college website — Oct 2025 |
| 🎓 Data Science Workshop | Ducat India — Model Training & Neural Networks — Nov 2025 |
| 🛠️ AI Training Program | Biyani Group of Colleges — Nov 2024 |
| 🌐 Google GDG DevFest | Attended GDG DevFest Jaipur |
| 💼 Software Developer Intern | Acmegrade, Bengaluru — ML with Python — Dec 2023 – Feb 2024 |
| Certificate | Issuer | Year | Core Skills |
|---|---|---|---|
| AI & Advanced IT Skills | IBM | Apr 2026 | LLMs, Jira Align, Enterprise AI |
| MUJ HackX 3.0 — International Hackathon | Manipal University Jaipur | Nov 2025 | Problem Solving, Team Engineering |
| Data Science Workshop | Ducat India | Nov 2025 | Model Training, Neural Networks |
| AI/ML — Chatbot Creation | Biyani Girls College | Oct 2025 | Python, Rasa, NLP |
| AI Training Program | Biyani Group of Colleges | Nov 2024 | AI Tooling, Public Speaking |
Languages
ML / DL / NLP
Data & Analysis
Tools & Platforms
learning:
- RAG Pipelines & Retrieval-Augmented Generation
- Fine-tuning LLMs with LoRA / QLoRA
- Transformer architectures from scratch
- React · Node.js · MongoDB full-stack track
building:
- AI-powered assistants with production-grade architecture
- NLP tools for real-world data problems
- Constraint-based optimization systems
exploring:
- Prompt engineering & LLM evaluation frameworks
- Vector databases (Pinecone, Weaviate, ChromaDB)
- Agent frameworks (LangChain, LlamaIndex)
- Open-source model deployment pipelines
open_to:
- AI/ML open-source collaborations
- Research internships in NLP / LLMs
- Hackathons & competitive AI challenges
- Entry-level AI/ML engineering roles
Working on something in AI/ML, NLP, or LLMs? Always open to collaborate.
"Continuous learning is the minimum requirement for success in any field."