Available for work

Tahir Khan.

Final Year Computer Science and Engineering Student

Computer Science undergraduate specializing in backend systems, machine learning, and scalable architectures. Practical experience architecting data-driven applications and developing predictive models. Strong foundation in data structures, algorithms, and system-level programming, with a track record of delivering robust solutions.

Bengaluru, India
tahir.kh2004@gmail.comLinkedInGitHub
+91 9591654570
Tahir Khan

Work Experience

Freelance Software Developer

Self-Employed

Jun 2025 — Present

Partnered with 2 distinct clients to architect, build, and deploy custom web applications tailored to their operational needs. Optimized system performance by implementing efficient caching layers and image compression algorithms.

Skills

C++PythonJavaScriptNode.jsFirebaseMongoDBSQLitePandasNumPyScikit-learnMatplotlibReact.jsHTML/CSSTailwindBootstrapGitOpenAI APIJinja2pdfkit

Projects

Predictive OS Bottleneck Detection

Architected a hybrid C++/Python system utilizing time-series analysis and workload-aware modeling to proactively detect OS bottlenecks. Developed predictive algorithms to identify and mitigate resource exhaustion before system performance degrades.

C++PythonTime-Series Analysis

Preschool Management Platform

Developed a full-stack dashboard for 100+ daily users, building a responsive frontend and structuring complex Firestore schemas for real-time announcements and payment tracking. Implemented robust role-based access control (RBAC) and optimized data retrieval to reduce query latency and ensure secure data isolation.

React.jsTailwind CSSNode.jsFirebase

Reddit Persona Generator (NLP)

Built an OSINT tool using PRAW and OpenAI models to scrape Reddit activity and generate detailed psychological and behavioral profiles. Designed a data pipeline outputting structured JSON with specific citations, text summaries, and dynamically styled PDF reports via Jinja2 and pdfkit.

PythonOpenAI APIPRAWJinja2pdfkitdotenv

Wallet Credit Scoring (ML)

Engineered an interpretable scoring model (300–1000 scale) for 1,000+ crypto wallets using KMeans to classify 6 distinct behavioral archetypes. Built a feature extraction pipeline processing on-chain transaction data to dynamically calculate base scores, bonuses, and penalties.

PythonPandasNumPyScikit-learnMatplotlibSeaborn

Education

B.Tech in Computer Science and Engineering (Cybersecurity)

Manipal Institute of Technology

Relevant Coursework: Operating Systems, Data Structures, Algorithms, Machine Learning, Cryptography

Aug 2022 — Present

Let's work together.

Open for new opportunities and high-impact collaborations.

Drop me an email