Rashmi Mallawaarachchi
AI EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Engineer with 1+ years in Generative AI & LLM Systems
AI/ML Engineer and Computer Engineering graduate specializing in applied machine learning, generative AI, and production deployment of LLM-powered systems. Hands-on experience designing and deploying end-to-end ML pipelines, including a production RAG-based tourism platform built with Pinecone, Google Gemini, and FastAPI on AWS. Strong foundation in deep learning (TensorFlow, PyTorch), classical ML (Scikit-learn), and supporting full-stack engineering. Seeking an entry-level AI/ML Engineer role where research-oriented thinking, model development skills, and engineering rigor can deliver measurable impact.
University of Ruhuna, Faculty of Engineering
BSc (Hons) · Computer Engineering
March 1, 2021 – January 1, 2026
BotCalm (Private) Limited
Full-Stack Developer Intern
August 1, 2024 – September 1, 2025
Matara, Southern Province, Sri Lanka
Face Mask Detection System (Group Project)
July 1, 2025 – October 1, 2025
Fine-tuned a MobileNet-based CNN on a dataset of labeled images for binary mask/no-mask classification, reaching 98% validation accuracy and 0.98 F1-score. Built the full preprocessing pipeline - image resizing (224×224), normalization, one-hot encoding, and augmentation (flip, rotation, zoom) to improve robustness under varied lighting and angles. Deployed the model as a real-time webcam application using OpenCV plus an interactive Streamlit UI for demos.
View ProjectAI-Aided Content Management System for Tourism (Final Year Project)
February 1, 2025 – November 1, 2025
Architected and deployed a production RAG pipeline (Google Gemini + Pinecone + LangChain) powering a multilingual tourism chatbot (7 languages) and intelligent trip planner, with persistent conversation memory. Designed and built FastAPI backend services exposing 20+ REST endpoints, integrating Firebase for transactional data and Pinecone for semantic retrieval. Implemented image-based content generation using vision-language embeddings, enabling automated tourist-attraction descriptions from user-uploaded photos. Built a location-aware recommendation service combining live geolocation with the OSRM routing engine to surface nearby destinations, services, and emergency contacts. Containerized the AI services with Docker and deployed on AWS EC2, establishing reproducible builds and zero-downtime updates across mobile and web clients.
View ProjectStock Price Prediction with LSTM (Individual)
June 1, 2024 – June 1, 2024
Built an end-to-end LSTM forecasting model on 4,427 days of historical stock data(5-day lookback window), achieving RMSE of 0.0167 and MAE of 0.0123 on held-out test data with an R2 of 0.962. Engineered supervised training data with sliding-window sequence construction; tuned sequence length, batch size, and network depth via systematic experiments. Implemented parallel training pipelines in both TensorFlow/Keras and PyTorch to benchmark framework differences in training time and convergence.
View ProjectFailing-Server Anomaly Detection (Individual)
January 1, 2024 – January 1, 2024
Built an unsupervised anomaly-detection system modeling normal server behavior (latency, throughput) with a multivariate Gaussian distribution, flagging failing servers in an 2-dimensional feature space. Selected the anomaly-probability threshold via cross-validation against an F1 objective on the validation set.
View ProjectCredit Card Approval Prediction (Individual)
January 1, 2024 – January 1, 2024
Built a binary classification pipeline on a 438,557-row credit dataset, comparing Logistic Regression and SVM. Addressed class imbalance with SMOTE oversampling, improving minority-class recall by 58% versus the imbalanced baseline. Performed correlation-based feature selection and outlier handling, reducing input dimensionality from 50+ features to key predictive features without loss of predictive power.
View ProjectGenerative AI in Practice: Advanced Insights and Operations
KodeCloud
June 1, 2026 – Present
Machine Learning Specialization
Coursera (DeepLearning.AI & Stanford)
June 1, 2026 – Present
MCP for Beginners
KodeCloud
June 1, 2026 – Present
Cultural Fit Analysis
The candidate's project portfolio demonstrates a diverse range of applications for AI, from financial forecasting to content management and anomaly detection, indicating adaptability and a broad interest in problem-solving. The academic and personal projects, combined with an internship, show initiative and a proactive approach to learning and applying new technologies. The target role of 'AI Engineer' aligns well with the candidate's demonstrated skills and project focus. However, the experience level is quite low, which might impact immediate cultural integration into a senior role without significant mentorship.
Soft Skills & Operational Fit
The candidate's project descriptions indicate an ability to work independently on complex problems (e.g., Stock Price Prediction, Anomaly Detection) and collaborate in team settings (e.g., AI-Aided Content Management System, Face Mask Detection). The internship experience highlights collaboration with cross-functional teams and adherence to clean code principles, suggesting good operational fit and teamwork potential. However, without direct assessment data, this remains an inference.