Software & AI Systems Engineer | High-Throughput APIs, Distributed Systems, Cloud Automation & Infrastructure Security

I am a Software and AI Systems Engineer specializing in high-throughput backend runtimes, distributed transactional architectures, and deterministic AI retrieval pipelines. My engineering approach centers on systems-level resilience: eliminating latency bottlenecks, mitigating state-mutation race conditions, and enforcing strict data-layer invariants. From architecting hybrid lexical/vector search engines with zero hallucinations to engineering 3-tier enterprise data warehouses processing over 1.05M records, I build fault-tolerant backends engineered for deterministic execution under load.

My practical experience spans concurrent async API microservices, distributed enterprise applications, automated quality harnesses, and infrastructure defense. I design across the entire stack—orchestrating containerized AWS cloud workflows, implementing database-level Row-Level Security, managing zero-data-loss disaster recovery pipelines, and enforcing strict operational SLAs through automated load testing and continuous deployment.

A decoupled, multi-tenant inventory control and operational state management platform engineered for strict data isolation and high-concurrency mutation workloads.

• Architecture & Storage: Engineered an Express.js and Node.js REST API with Mongoose ODM, utilizing compound B-tree indexes ({ ownerUserId: 1, purchased: 1 }) and resilient multi-stage database bootstrapping with automatic failover from persistent WiredTiger disk storage to in-memory instances under lock contention.
• Ownership Enforcement: Implemented stateless JWT Bearer token authentication with multi-tenant object-level access verification, eliminating unauthorized cross-tenant mutations.
• Performance & Verification: Validated under concurrent stress using Grafana k6 pipelines to guarantee SLAs ($p(95) < 200\text{ms}$, failure rate $< 1%$ across 100 concurrent VUs) alongside Newman/Postman automated collection regression suites.
• Tech Stack: Node.js | Express.js | MongoDB (WiredTiger) | Jest | Grafana k6 | Postman / Newman
• Source: [Institutional Repository — Access Available by Request]

An asynchronous scriptural intelligence engine combining a Next.js 15 App Router frontend with a high-throughput FastAPI backend running a hybrid lexical/semantic retrieval pipeline across 5,034 canonical commentary pages.

• Hybrid Retrieval Pipeline: Architected an in-memory lexical BM25Okapi search engine (Robertson-Spärck Jones IDF with document-length normalization) coupled with Reciprocal Rank Fusion (k=60) across 5 weighted datasets (W=3.0 down to W=1.0) indexing 649 unique shlokas and 13,226 vocabulary tokens.
• Multi-Model LLM Ensemble & Arbitration: Asynchronous parallel fan-out across multiple LLM providers (Groq, OpenRouter) arbitrated by an LLM judge model with post-generation ground-truth citation verification.
• Resilience & Fault Tolerance: Implemented an asynchronous 3-state circuit breaker (CLOSED, OPEN, HALF-OPEN) with a 5-failure threshold and 30s cooldown to isolate upstream rate limits, serving responses via chunked Server-Sent Events (SSE) with buffer suppression (X-Accel-Buffering: no).
• Container Security & Persistence: Multi-stage Docker deployment running under a dedicated non-root execution user (appuser, UID 10001) backed by PostgreSQL session storage with 18-day TTL tokens.
• Verification & CI/CD: Hardened with strict Pydantic v2 input boundary validation, an automated 39-test Pytest verification suite covering citation scrubbers and wire protocols (100% pass rate), and an automated GitHub Actions CI/CD pipeline.
• Tech Stack: FastAPI | Python 3.12 | Next.js 15 | React 19 | PostgreSQL | Redis | Docker | Pytest | Groq API
• Source:

An applied machine learning operations (MLOps) and intelligence portfolio featuring automated model lifecycle management, statistical data drift monitoring, and deep learning pipelines.

• Continuous Audio MLOps Pipeline: Trained and benchmarked acoustic classification models (Logistic Regression, Decision Trees, K-Nearest Neighbors) across 232,725 historical Spotify tracks with automated 80/20 train/test splitting and StandardScaler serialization parity.
• MLflow Tracking & Model Registry: Integrated automated MLflow experiment tracking logging hyperparameters, confusion matrices, and model accuracy, with an automated Model Registry gateway registering winning candidate models to production stages.
• Statistical Data Drift Monitoring: Engineered continuous covariate shift detection with Evidently AI and SciPy, executing Two-Sample Kolmogorov-Smirnov (KS-test) and Wasserstein distance calculations to export automated HTML and JSON drift diagnostic reports.
• Kubeflow Pipelines (KFP v2) DAG: Compiled a declarative 5-stage Kubeflow Pipeline DAG (audio_pipeline_dag.yaml) orchestrating data validation, model retraining, MLflow serialization, and automated drift alerting.
• Cryptographic Verification Ledger: Developed SkillChainLedger, an append-only SHA-256 cryptographic blockchain ledger in Python, coupled with a reactive Streamlit analytics dashboard and deep learning classification modules (IMDB NLP & MNIST neural networks).
• Tech Stack: Python 3.13 | MLflow | Kubeflow Pipelines (KFP v2) | Evidently AI | Scikit-Learn | TensorFlow / Keras | Streamlit | SciPy | Pandas | Pytest
• Source:

A production-grade cybersecurity engineering suite modeling real-world attack surfaces, executing penetration tests, automating disaster recovery, and enforcing host-hardening baselines across segmented subnets.

• Offensive & Defensive Security Operations: Executed automated DAST fuzzing on Docker microservices via OWASP ZAP reverse-proxies, mapped attack surfaces with Greenbone GVM across full IANA ranges, and executed black-box Metasploit penetration runs against vulnerable Linux daemons.
• Continuous Disaster Recovery: Architected an automated zero-data-loss PostgreSQL backup pipeline streaming pg_dump directly into an encrypted, content-addressable Restic repository via non-interactive nightly Cron jobs (AES-256 at rest, SHA-256 deduplication).
• Host Hardening & Auditing: Automated declarative host hardening using Ansible Core to enforce default-deny UFW perimeters (ports 22/8080), auto-generate X.509 TLS PKI certificates, and establish AIDE SHA-256 tripwires across critical paths (/etc/passwd, /etc/shadow, /etc/ssh/sshd_config).
• Threat Modeling & Governance: Authored enterprise STRIDE threat models for autonomous transport architectures (mTLS, HSM, CAN Bus isolation) and drafted PHIPA-compliant healthcare IAM/MDM governance frameworks.
• Tech Stack: Kali Linux | Ubuntu Server | Ansible Core | PostgreSQL | Restic | OWASP ZAP | OpenVAS / GVM | Metasploit | Docker
• Source:

A cross-platform professional consulting marketplace featuring vector-based semantic matchmaking, spatial distance filtering, and atomic contract lifecycle workflows.

• Mobile Architecture & Rendering: Solely architected the cross-platform client using React Native (Expo SDK 54) and TypeScript, integrating TanStack Query for cache invalidation and NativeWind UI to eradicate layout shifts.
• Semantic & Spatial Discovery: Integrated client discovery with Supabase PostgreSQL RPCs, combining 384-dimensional pgvector embeddings via HNSW cosine indexing (vector_cosine_ops) with PostGIS GiST spatial queries (ST_DWithin) and composite scoring algorithms.
• Real-Time Communications & Security: Built a duplex messaging layer backed by Supabase WebSockets, PL/pgSQL database triggers, and database-level idempotent event keys under 100% Row-Level Security (RLS) via isolated SECURITY DEFINER procedures.
• Tech Stack: React Native (Expo SDK 54) | TypeScript | Supabase | PostgreSQL | pgvector | PostGIS | TanStack Query
• Source: [Proprietary Architecture — Source Protected under Signed NDA]

A suite of high-availability enterprise applications engineered with strict design patterns and distributed transaction processing capabilities.

• Jakarta EE Enterprise Platform (ACMEMedical): Architected a multi-tier clinical management platform on Java 21 and Payara Server using JAX-RS, stateless EJBs, and JPA/Hibernate with optimistic locking (@Version), RFC 7617 HTTP Basic authentication, and Soteria PBKDF2 credential hashing (2048 iterations).
• Transit Fleet Operations (PTFMS): Built an enterprise transit coordination platform using Java 8 / Servlets, JDBC PreparedStatement pools, and GoF patterns (Strategy for polymorphic propulsion fuel tracking, Adapter for external GPS feeds, and Observer for maintenance dispatching).
• Enterprise Persistence: Engineered normalized schemas across MySQL instances with connection pooling, declarative role-based access control, and JUnit 5 / Jersey Client integration suites.
• Tech Stack: Java 21 / 8 | Jakarta EE | EJB | JPA / Hibernate | Payara | Servlets | JDBC | MySQL
• Source:

A decision-support banking data warehouse modeled on a 3-tier Kimball analytical pattern to ingest, model, and analyze transactional ledgers and forecast credit risk.

• High-Volume ETL Architecture: Engineered SSIS memory-buffered pipelines with upstream physical sort contracts and cascading merge joins to extract, cleanse, and transform $1.05\text{M}+$ financial records and regional socio-economic indicators.
• Predictive Risk Analytics: Authored multi-level Transact-SQL CTEs, rolling aggregations, and window functions to compute liquidity depletion velocity flags ($>5$ withdrawals $> $500$), forecasting loan defaults 2–3 months prior to delinquency.
• Dimensional Modeling: Designed star/snowflake schemas in SQL Server with clustered indexing, categorical domain mapping, and interactive Power BI analytical dashboards.
• Tech Stack: Microsoft SQL Server | T-SQL | SSIS | SSMS | Power BI
• Source:

An enterprise-grade test automation and regression framework designed to evaluate single-page web applications against complex shopping and checkout flows.

• DOM Synchronization Engine: Engineered a Page Object Model (POM) test architecture in Python and Selenium WebDriver, injecting custom JavaScript into React’s internal _valueTracker to dispatch synthetic bubbling input and change events and eliminate headless synchronization race conditions.
• Dual Automation Engine: Implemented 22 parameterized Pytest scenarios and 14 keyword-driven Robot Framework specifications with zero implicit waits and explicit condition polling.
• Continuous Integration Pipelines: Configured dual execution pipelines across Jenkins (declarative Jenkinsfile archiving JUnit XML trends) and GitHub Actions with automated failure screenshot capture and interactive HTML test reporting.
• Tech Stack: Python | Selenium WebDriver | Robot Framework | Pytest | Jenkins | GitHub Actions
• Source:

• Algonquin College (Ottawa, ON)
  • Advanced Diploma in Computer Programming and Analysis (AAL 01–06)
  • Cumulative GPA: 3.52 / 4.00 | Multi-Term Dean's Honours List
• Sir Bhavsinhji Polytechnic Institute (Gujarat, India)
  • Diploma in Information Technology
  • Cumulative GPA: 7.91 / 10.0 CGPA (First Class with Distinction)

• AWS Certified Cloud Practitioner (CLF-C02) — In Progress
• AWS Educate – Introduction to Cloud 101
• Microsoft – Describe Cloud Computing Principles
• LinkedIn Learning – Software Architecture Patterns