Senior Full-Stack Engineer (Product & Platform Focused)
I build product-ready systems that stay reliable under real traffic.
End-to-end ownership: UI, APIs, data, deployment. Event-driven systems, reliability patterns, production readiness.
Customer-facing workflows, stakeholder collaboration, measurable improvements.
Latency -30% on critical endpointsDesigned and delivered backend services that directly powered customer-facing web workflows used by thousands of users.
Event ledger
Event Bus
Event-driven flow
Watch idempotency, retries, outbox, and DLQ kick in.
Systems playground
Trace the flow, spot the safeguards.
Ctrl + K for command palette
Architecture Mode turns the whole site into a diagram-first view and highlights where each project lives.
Event-driven map
Service mesh in miniature
API gateways, compute, storage, and queues stitched with observability. Projects attach right where real incidents happen.
Core services Projects anchored to the flow Selected node
Focused node
API Gateway
Auth, rate limits, and idempotency tokens at the edge.
Reliability scorecard
How the systems stay calm under pressure
Patterns I ship by default. Hover or focus each row for why it exists.
SLO target
99.9%Experience
Owning services end-to-end in production.
Expandable timeline
Fixfirst - Backend Engineer (Product-Facing Systems) Germany (remote) Latency -30% January 2024 - November 2025
Germany (remote)
Latency -30%
- Designed and delivered backend services that directly powered customer-facing web workflows used by thousands of users.
- Owned API contracts end-to-end, collaborating closely with frontend teams to support evolving product features.
- Translated product requirements into scalable services and data models used across multiple user journeys.
- Improved customer experience by reducing latency by 30% on critical endpoints and failure rates across critical paths.
- Implemented production-ready reliability patterns, security controls, and CI/CD pipelines to support continuous delivery.
What I shipped
- Customer-facing workflow services and APIs
- Shared data models across user journeys
- Reliability, security, CI/CD for production delivery
How it ran in prod
- Latency and failure-rate gains on critical paths
- Reliability patterns and safeguards
- CI/CD pipelines for continuous releases
SMSami - Full-Stack Developer Pakistan January 2021 - Aug 2023
Pakistan
- Owned full-stack features from UI implementation to backend logic and production deployment.
- Built and maintained customer-facing dashboards and internal tools used by real users.
- Delivered end-to-end solutions covering frontend interfaces, APIs, data storage, and cloud deployment.
- Worked directly with stakeholders to refine requirements, iterate on features, and respond to user feedback.
- Mentored junior developers and contributed to code reviews and architectural discussions.
What I shipped
- Customer-facing dashboards and internal tools
- End-to-end features across UI, APIs, data
- Stakeholder-driven feature iterations
How it ran in prod
- Production deployments across frontend and backend
- Operational support for user-facing workflows
- Code review and architecture standards
AYS Electronics - Junior Full-Stack Developer Pakistan June 2019 - Dec 2020
Pakistan
- Implemented frontend features and backend endpoints as part of live production systems.
- Worked on real customer requirements, bug fixes, and incremental feature delivery.
- Gained hands-on experience across UI development, backend services, and database integrations.
What I shipped
- Frontend features and backend endpoints
- Incremental features and bug fixes
- Database integrations for production workflows
How it ran in prod
- Production support and issue fixes
- Cross-stack delivery from UI to backend
- UI, backend, database integrations
Projects
Mini case studies with interactive demos.
Each project shows the problem, the engineering approach, and the operational outcome.
Sentinel Incident Platform
Incident workflow coordination with reliable event processing.
Architecture snapshot
Webhook intake flows through retries, DLQ, and replay tooling.
Problem
Incident workflows break when payment webhooks fail or arrive twice, making recovery slow and hard to audit.
Approach
- Designed event-driven intake with correlation IDs and idempotency at the edge.
- Modeled incident state changes with an outbox and DLQ for replayable recovery.
- Instrumented traces, metrics, and logs to tie failures to user impact.
Outcome
- Delivered resilient incident workflows with safe retries and replay tooling.
- Made failures traceable across services for faster triage.
How users interact with it
Operators trigger incidents, review retries, and replay failed events from a shared console.
What breaks first: downstream webhook providers stall, pushing retries into the DLQ and stretching recovery timelines.
Tradeoffs
- Chose: event-driven intake with idempotency + DLQ replay.
- Rejected: synchronous processing with manual incident recovery.
- Why: needed auditability and safe retries under real traffic spikes.
Incident console
Status: Idle
Latency
120 ms
Error rate
2.0%
No incidents yet. Trigger an event to start.
Correlation ID: cor-bb0b07 - Idempotency key: idem-44e189
Key engineering decisions
- Idempotency keys gate webhook retries to prevent double writes.
- Outbox events keep data changes and emissions consistent.
- DLQ + replay tooling isolates failures without losing events.
- Correlation IDs flow through logs, traces, and alerts.
Network Anomaly Detection
Detect abnormal behavior from logs without labeled data.
Architecture snapshot
Feature scoring feeds the model and raises alerts at tuned thresholds.
Problem
Ops teams need to detect abnormal network behavior from logs without labeled incident data.
Approach
- Built a feature pipeline and synthetic evaluation set to test sensitivity.
- Compared Isolation Forest vs One-Class SVM for unsupervised detection.
- Exposed contamination and threshold controls to make tradeoffs explicit.
Outcome
- Delivered actionable anomaly signals for monitoring use cases.
- Made model tradeoffs transparent for stakeholders.
How users interact with it
Analysts tune thresholds, compare models, and review highlighted anomalies in a time-series view.
What breaks first: contamination drift pushes false positives before thresholds are recalibrated.
Tradeoffs
- Chose: transparent model toggles with operator-tuned thresholds.
- Rejected: black-box auto-thresholding without operator context.
- Why: ops teams needed explainable tradeoffs between noise and misses.
Toggle between models and tune contamination vs threshold to see precision/recall tradeoffs.
Predicted Actual anomalies
Precision
100%
Recall
50%
Isolation Forest expects a contamination ratio and flags outliers by density splits.
Key engineering decisions
- Unsupervised models avoid labeled data bottlenecks.
- Standardized distance scoring keeps thresholds consistent across runs.
- UI surfaces precision and recall to guide operations teams.
Capability map
Focused clusters instead of a wall of buzzwords.
API & Data Modeling
- REST design
- Schema versioning
- Postgres
- DynamoDB
Async & Messaging
- SQS/SNS
- EventBridge
- Kafka-style patterns
- Queue workers
Reliability Patterns
- Retries + backoff
- DLQ
- Outbox
- Idempotency
Security Fundamentals
- OAuth2
- RBAC
- Audit trails
- Secrets management
Observability & CI/CD
- Tracing
- Metrics + SLOs
- GitHub Actions
- Canary deploys
AWS Serverless
- Lambda
- API Gateway
- DynamoDB/RDS
- SQS/SNS/EventBridge
- Cognito
- CloudWatch/X-Ray
Certifications & Education
Security and data science grounded in engineering.
Certified Ethical Hacker (CEH)
EC-Council
Data Science & Machine Learning
Professional Certification
BEng in Computer Science
University of Engineering & Technology
Open to roles
Open to roles: Full-Stack / Backend (Product-facing)
London / hybrid / remote
Contact
Let's talk about product-facing systems.
Copy the email or open your mail client.
Email
info@shayanmudassar.co.uk