Hi, I'm Walter Banda
Software Engineer — Full Stack, AI & Agent Systems, and Cloud. Taking products from data model to production.
Powered by the Best Tech Stacks and Frameworks
What Sets Me Apart
A track record of delivering impactful solutions across the full technology stack
AI & Agent Systems
Multi-agent LLM pipelines, RAG agents on AWS Bedrock, generative image and video production with Gemini and Veo, and a QLoRA fine-tune of a language model for Shona.
AWS Certified
AWS Certified Cloud Practitioner with production experience across Lambda, S3, EKS, Kinesis, App Runner and Bedrock — building and shipping real workloads, not just labs.
KCNA Certified
Kubernetes and Cloud Native Associate. Services containerised with Docker and Helm, orchestrated on Kubernetes, and released through Argo CD pipelines.
Audio & Signal Processing
Built a Shazam-style audio fingerprinting engine that detects adverts inside live radio broadcasts, with a faster-whisper stage that catches spoken keywords too.
DevOps & Observability
CI/CD with GitLab CI and GitHub Actions, services instrumented with Prometheus and Grafana, JWT auth and RBAC enforced across APIs.
BSc (Hons) Computer Science
Second Class, Upper Division — Africa University. IITPSA Professional Member. Strong foundation in algorithms, distributed computing and system design.
Full Stack Capabilities
End-to-end expertise from pixel-perfect frontends to robust backend systems and cloud infrastructure
Frontend Development
User Interface LayerStandalone components, signals, RxJS reactive programming, D3 dashboards
SSR/SSG applications, admin platforms and marketing sites in production
Cross-platform mobile apps with a single codebase for iOS & Android
Backend Development
Server LayerREST and GraphQL APIs, WebSockets, RabbitMQ-backed microservices
Layered handler → service → repository backends with real-time WebSocket messaging
Async services for audio processing, ML pipelines and agent systems
Database Management
Data LayerNoSQL document storage, aggregation pipelines, indexing strategies
Relational database design, query optimization, performance tuning
In-memory caching, session management, real-time data stores
Cloud & DevOps
Infrastructure LayerLambda, S3, EKS, Kinesis, App Runner and Bedrock in production
Containerised microservices released through Argo CD pipelines
GitLab CI, GitHub Actions, Nginx, Prometheus & Grafana observability
Featured Projects
Real-world applications I've built and contributed to
Tesha
Founder & Lead Engineer · tesha.co.zwMarketplace connecting Zimbabwean informal workers with paid odd jobs. When mobile-data costs blocked adoption, I re-architected the service to run entirely inside WhatsApp — putting it within reach of users who can't afford a conventional app.
Tesha Creative Studio
AI Asset Generation & Social PublishingA prompt engine turns stored brand intelligence into finished marketing imagery and video via Gemini and Veo — no prompt writing. Posts authored once as blogs are AI-rewritten into platform-native captions and published to X, Facebook, Instagram, LinkedIn, Threads and TikTok. Live ads have shipped through it.
Dynaton Ad Exchange
Real-Time Bidding PlatformReal-time bidding exchange serving display, SMS and mobile campaigns — campaign moderation, tracking pixels, proof-of-flight verification, publisher payout reconciliation and CDN asset delivery.
AdHive
adhive.africa · Influencer MarketingInfluencer-marketing platform with a layered Go backend, JWT auth and real-time WebSocket messaging. I built the Paynow Zimbabwe payment integration — EcoCash, OneMoney and card — with SHA-512 signing and webhook validation.
Lead Discovery & AI Outreach
Automated Lead GenerationScrapers sweep social platforms for small businesses, score them with an informality index that surfaces unregistered traders, then a RAG agent on AWS Bedrock qualifies each lead and a WhatsApp agent opens the conversation and books the call.
Broadcast Advert Detection
Audio Fingerprinting EngineShazam-style fingerprinting engine that detects client adverts inside live radio streams — spectrogram peak constellation hashing matched by time-offset histogram, with a faster-whisper stage that catches spoken keyword mentions too.
Shona Language Model
Low-Resource LLM Fine-TuningEnd-to-end fine-tuning pipeline for a low-resource African language: a scraped Shona corpus, ~40,000 instruction pairs generated by a synthesiser-and-critic loop, QLoRA fine-tuning of a 4B model, quantised to GGUF for local inference.
Brands IQ
Brand Monitoring & AnalyticsBrand monitoring and competitor analytics across Reddit, X, YouTube, Instagram, TikTok and Facebook — a scraping pipeline feeding multi-provider sentiment analysis through LangChain and AWS Comprehend, with multilingual language detection.
DICOMM Electricity
Offline-First Utility PlatformOffline-first meter-reading and consumption platform for a prepaid electricity business — writes queue locally and drain on reconnect, with 30-day analytics, quotations and scheduled Google Drive backup. Built to WCAG AA.
Zim Legal Aid
AI-Powered Legal AssistantMultilingual legal aid platform helping Zimbabweans understand their rights, build cases, and generate legal documents in English, Shona, and Ndebele — powered by AWS Bedrock RAG.
FarmPulse
Open Market IntelligenceAgricultural market intelligence platform for Zimbabwean farmers — featuring real-time commodity price analysis, AI-powered crop advice, comparison tools, and a community forum.
Virtual Science Labs
virtualsciencelabs.academyVirtual chemistry laboratory platform for ZIMSEC students — enabling interactive experiments and hands-on learning aligned with the Zimbabwean curriculum.
Zvigaba
Voice-First Financial AdvisorVoice-first AI financial advisor for Zimbabwe's informal traders — record transactions by voice in English or Shona, get real-time financial analytics, and receive smart business advice powered by Amazon Nova.
Hi, I'm Walter
I'm a software engineer with five years of experience taking products from data model to production — Angular, Next.js, Nuxt and Ionic on the front end; Node.js, Go, Python and Java behind them; AWS, GCP and Firebase underneath.
I work on a real-time bidding ad exchange and the products around it, and my recent work is concentrated in applied AI: multi-agent LLM pipelines, generative image and video production, RAG agents on AWS Bedrock, an audio-fingerprinting engine that detects adverts in live radio, and a QLoRA fine-tune of a language model for Shona.
I hold a BSc (Hons) in Computer Science (Upper Second), am an AWS Certified Cloud Practitioner and KCNA certified, and I'm used to owning a service end to end in small distributed teams.
Let's Build Something Amazing
Whether you're looking for a new project, partnership, or have an innovative idea to discuss, I'm here to connect and make things happen.