Akhil Salla | AI & Software Engineer

About Me

Career Objective

AI/ML engineer with strong expertise in Generative AI, Large Language Models, NLP, and Computer Vision. Skilled in Python, MLOps, and automation, with proven ability to design, deploy, and optimize intelligent solutions.

Passionate about transforming ideas into impactful AI applications, I combine technical depth with practical problem-solving to deliver scalable, real-world innovations. Seeking opportunities to apply AI/ML expertise to impactful real-world applications.

Experience

Software Engineer

Mar 2023 – Sep 2023

Nexquare

Developed the School Fee API, streamlining revenue tracking and reducing administrative costs by 20%.
Built an Automatic Label Encoder to reuse mapped labels, cutting development downtime by 90%.
Designed a multi-user API for managing student syllabi, doubling content management productivity.

Data Scientist

Aug 2022 – Oct 2022

21K School

Analyzed 5TB of administrative data for lead generation and behavior prediction, increasing leads by 30%.
Performed feature engineering, selection, and developed high-accuracy ML models to automate insights.
Delivered tools that automated report generation for school administration.

Machine Learning Mentor

Apr 2021 – Aug 2022

EpsilonPi, Hyderabad, India

Conducted Python and ML bootcamps, mentoring students on 30+ MVPs.
Organized a 24-hour hackathon with sponsors like Microsoft.
Provided personalized guidance on ML projects.

Machine Learning Intern

Jul 2021 – Apr 2022

CodVis, Hyderabad, India

Led a 6-member ML team to deliver 5 customer-focused features within 3 months.
Coordinated cross-functional teams to integrate AI into products, reducing development time by 25%.
Developed data visualization tools using Python, Django, and ML techniques.

Education

Bachelor of Technology, Computer Science and Engineering

2018 – 2022

Vignana Bharathi Institute of Technology, Hyderabad, India

Technical Skills

Programming & Scripting

Python C++ SQL Bash

Frameworks & Libraries

TensorFlow PyTorch Scikit-learn LangChain Hugging Face Transformers Pandas NumPy OpenCV FastAPI Django Seaborn Matplotlib Plotly

Generative AI & LLMs

Prompt Engineering Retrieval-Augmented Generation (RAG) Conversational Memory LangChain Agents & Toolkits Vector Stores (FAISS, Chroma) LLM Caching & Tracing LangServe OpenAI API (GPT-3.5/4) Groq API Ollama (Local LLMs) Hugging Face (Spaces, Transformers)

Machine Learning & AI

Supervised & Unsupervised Learning Predictive Modeling Feature Engineering Model Evaluation & Tuning Deep Learning Transfer Learning Computer Vision Natural Language Processing Named Entity Recognition Sentiment Analysis Topic Modeling

APIs & Web Technologies

REST API Development & Testing LLM API Integration Postman Streamlit (AI Dashboards)

Tools

JupyterLab VS Code Git/GitHub Docker Conda Linux

Cloud & Deployment

FastAPI Streamlit Sharing LangServe Hugging Face Spaces AWS EC2 & S3 (basic) Model Serialization (Pickle)

Data & Analysis

Data Wrangling Statistical Analysis Exploratory Data Analysis Data Visualization (Seaborn, Matplotlib, Plotly) Time Series (basic) SQL Data Extraction

Languages

English, Hindi, Telugu

Projects

AI Search Engine With Tools And Agents

An AI-powered search engine built with LangChain, Groq, and Streamlit, integrating tools like Wikipedia, Arxiv, and DuckDuckGo to provide intelligent, multi-source answers with agent-based reasoning.

LangChain Groq Streamlit Wikipedia / Arxiv / DuckDuckGo APIs FAISS

Talk With Docs

A RAG-based chatbot that lets users upload documents and interact with them via natural language. Combines embeddings, vector search, and LLMs with chat history to maintain contextual, intelligent responses.

LangChain SentenceTransformers Streamlit OpenAI / Groq FAISS

Multi Provider Q&A Chatbot

A Q&A chatbot integrating multiple LLM providers like OpenAI, Groq, and HuggingFace. Enables real-time response comparison, dynamic provider switching, and robust fallback mechanisms through a unified conversational interface.

LangChain DuckDuckGo / Wikipedia Tools Streamlit OpenAI / Groq Chat Memory

AWS DynamoDB RCU Provisioning Optimization

Built a machine learning pipeline to classify DynamoDB RCU provisioning states (over/under/balanced) using historical usage data, time-based features, and Random Forests for predictive infrastructure scaling

Pandas & NumPy Matplotlib & Seaborn Scikit-learn Random Forest Classifier Jupyter Notebook

Fine-Grained Image Classification (CV)

Employs deep learning (ResNet, EfficientNet) to classify images into highly similar subcategories. Uses transfer learning. Evaluated by Top-1/Top-5 accuracy.

Deep Learning Computer Vision Transfer Learning ResNet

Voice to Intent Classification (DL)

Converts speech to text, then classifies user intent using NLP models (CNNs, transformers) to map voice commands to actions. Evaluated by accuracy/intent recognition rate.

Deep Learning NLP Speech Recognition Transformers

Image Generation with GAN (CV)

Uses Generative Adversarial Networks (GANs) to generate realistic images. Focuses on improving image quality/diversity. Evaluated using Inception Score/FID.

Deep Learning Computer Vision GAN Generative AI

Toxic Comment Detection (Deep Learning)

Leverages deep learning (LSTM, BERT) to identify toxic comments. Processes text data for contextual understanding. Evaluated using AUC-ROC or precision-recall.

Deep Learning NLP LSTM BERT

Customer Churn Prediction (Deep Learning)

Developed a customer churn prediction model using bank customer data. Applied data preprocessing, feature selection, and machine learning techniques to identify high-risk customers and reduce customer attrition.

Pandas & NumPy Matplotlib & Seaborn Scikit-learn Logistic Regression / Random Forest / XGBoost

Countries Life Expectany Prediction

Built a life expectancy prediction pipeline using cleaned datasets, linear regression modeling, and a Flask-based web interface—enabling live predictions through a deployed model and intuitive REST API.

Pandas & NumPy Pickle Scikit-learn Flask EDA