Headline

Sr. AI/ML Engineer || Agentic AI || Voice Bot || Chat Bot || Data science || PHP

About

Vision IAS leverages advanced AI systems to empower UPSC aspirants through innovative tools such as an AI-powered mock interview platform. This project integrates dynamic question generation and feedback systems using Python, GPT-4, and ChromaDB, improving student preparation. Designed speech-to-text and text-to-speech features using React, enabling intuitive interactions. Graduated with a BE in ECE from Maharshi Dayanand University, their work reflects a strong foundation in engineering principles applied to AI and education. Motivated by a passion for streamlining student support, they contribute to building scalable AI solutions that enhance user experiences while reducing operational inefficiencies.

Location

Noida

Work experience

Senior AI/ML Engineer at University Living, Nov 2025 - Present · 8 mos, Noida, Uttar Pradesh, India, 1. AI-Powered Sales Calling Bot for Student Abroad Services Objective: To build an end-to-end AI-driven outbound calling system that automatically connects with student leads, understands their requirements, and provides accurate responses regarding accommodation, flights, and other overseas support services. Achievements: Designed and developed a fully automated AI voice agent capable of natural, human-like conversations. Implemented real-time speech-to-text and text-to-speech pipelines enabling smooth two-way communication. Integrated dynamic conversation flows to answer student queries about accommodations, flight options, and documentation support. Reduced manual calling effort by 70% by automating initial lead qualification. Improved lead engagement and conversion rate by identifying students with genuine interest before passing them to the staff. Tech Stack: Python, FastAPI, WebSockets, Redis, GPT-based LLM, Speech-to-Text (Whisper/Deepgram), Text-to-Speech (ElevenLabs/Google TTS), Twilio, Exotel, Postgress DB, Docker, Cloud Deployment (AWS), Streamlit, Audio Model (openai / eleven Lab) 2: Call Conversation Analysis & Lead Scoring System Objective: To create an automated system that analyzes recorded AI calls and identifies which leads show positive intent, helping the sales team prioritize high-quality prospects. Achievements: ● Built a conversation analytics engine to evaluate intent, sentiment, clarity of requirements, and conversion probability. ● Developed a scoring model that categorizes calls as positive, neutral, or negative based on predefined business metrics. ● Automated daily lead scoring, enabling the team to focus on high-intent students only. ● Improved follow-up efficiency and reduced staff workload by 50%. ● Provided structured call summary and recommended next actions for each lead. Tech Stack: Python, GPT/LLM-based Analysis, DeepInfra for audio to text, Sentiment Analysis, FastAPI, PostgreSQL/MySQL, Docker, SQLite3 for Queue

AI Engineer at Vision IAS, Mar 2022 - Feb 2026 · 4 yrs, Noida, Uttar Pradesh, India, AI-Powered Mock Interview System for UPSC Aspirants Objective : Build a system that simulates a job interview environment, where students can practice and get feedback on their answers performance. Key Features Question Generation: Dynamically generates interview questions based on academic background, location, current affairs. Feedback System: Provides detailed feedback on responses, and suggests areas of improvement. Tech Stack: Python, GPT-4, ChromaDB, React (react-speech-recognition) for speech to text convert, also react use for text to speech. AI Counsellor for Educational Website • Tech Stack: Python, GPT-4, ChromaDB, NLP, Automation Tools • Objective: Streamline AI-driven student support while reducing the support team from 5-7 people to 1- Developed Large Language Model (LLM) from Scratch • Objective: 124 million parameter LLM for versatile text generation. • Technologies Used: Python, Transformer Architecture, PyTorch. Fine-tuning LLM for UPSC-MCQ Generation • Technologies Used: LLM (base model), PEFT, Tokenizer, Hyperparameter, Dataset, prompts • Objective: Reducing Teacher Dependency in Generating UPSC-Level MCQ • Achievements: Achieved 60% cost reduction and generated over 2000 high-quality MCQs. Enhanced relevance and accuracy, level of difficult and quality of generated questions are 80 -90% . Estimated time savings of 150+ hours for question creators. Question Answering System using RAG and ChromaDB • Objective: Develop a QA system to extract accurate answers from PDF databases. • Technologies Used: ChromaDB, LangChain, RAG, PDF parsing libraries. • Achievements: saved 50% of time on manual searches. Student Performance Prediction • Objective: Develop a system to predict student academic performance based on past grades, attendance, and behavioral data. • Tech Stack: Regression algorithms (e.g., Linear Regression, Random Forests), Hyperparameter optimizer. • Impact: Helps educators identify students who may need additional support.

Full-stack Developer (LAMP) at Vision IAS, Oct 2020 - Feb 2022 · 1 yr 5 mos, Delhi, India, • Core PHP and CodeIgniter Framework. • Designed and implemented a comprehensive student testing feature, enabling students to take assessments, track attempts, and thoroughly analyze their test results for enhanced learning insights. • Developed and optimized robust code for the admin portal, ensuring seamless query execution and data processing. • Managed MySQL data processing tasks, implementing advanced query optimization techniques for enhanced system performance remove the JOIN and sub-query. • Engineered and maintained high-performance web applications using PHP, HTML, CSS, JavaScript, and MySQL, Grocery CRUD. • Optimized data processing from MySQL servers, efficiently managing millions of entries and improving system performance.

PHP Developer at Zambo Technology, Feb 2019 - Sep 2020 · 1 yr 8 mos, Lucknow, Uttar Pradesh, India, • Core PHP and Codeigniter Framework. • Spearheaded secure banking API integrations (ICICI BANK, RAZORPAY), revolutionizing transaction processes and fortifying data protection measures. • Implemented robust security protocols including: o SQL Injection Prevention o CORS (Cross-Origin Resource Sharing) o JWT (JSON Web Tokens) authentication o XSS (Cross-Site Scripting) protection o Cross-Site Request Forgery (CSRF) safeguards • Integrated AePS (Aadhaar Enabled Payment Service) with Mantra and Morpho devices, expanding payment options and enhancing user accessibility. • Pioneered the development of cutting-edge features for financial transactions, showcasing expertise in complex data handling and real-time processing.

Education

Bachelor of Engineering - BE, ECE, Maharshi Dayanand University Rohtak