Masih
Pourmoalem

Implemented a system that turns a user-uploaded photo into a fantasy/mythological-style digital artwork, mints it as an NFT on Polygon, and completes purchase and transfer to the user’s MetaMask wallet. The flow includes server-side or API-driven style transfer/generation, NFT minting and metadata on Polygon, and a payment step so funds support the UN Sustainable Development Goals (SDGs). Technical scope: image ingestion and processing, generative or style-transfer output, smart-contract or platform integration for minting on Polygon, payment handling, and wallet delivery (MetaMask). Result: a working Web3 product for UN SDG fundraising via user-generated NFT art on Polygon.

Implemented end-to-end data collection and extraction systems for multiple clients: custom crawlers and scrapers, plus integration with specialized extraction tools where needed, depending on source type (web, APIs, directories). Pipelines support ingestion, normalization, and storage of large volumes of data for market research, sales lead generation, competitor and pricing analysis, and market/user behavior monitoring—enabling data-driven decisions at low marginal cost. Deployed across several companies. Result: new sales channels built on extracted data and, in one case, a large-scale market research program powered by the same pipeline.

Implemented a network management and monitoring system that gives the client visibility and control across the infrastructure: traffic analysis and flow data for attack and intrusion detection; bandwidth management and policy enforcement; and detection of suspicious behavior, including IP spoofing and anomalous patterns. The system supports centralized monitoring, alerting, and mapping so admins can identify threats, manage capacity, and respond to incidents. Result: improved security posture, controlled bandwidth use, and earlier detection of attacks and abuse.

Designed a NAS for an architecture team as a centralized shared drive (SMB/NFS) for fast data exchange. NAS-side I/O and caching (CPU/RAM on appliance) used to limit redundant network traffic and bandwidth saturation. Concurrent access supported so multiple users could work on the same project without speed loss (RAID, optional SSD cache, adequate network). Outcome: efficient, low-latency shared storage for collaborative 3D and design workflows.

Implemented a NAS storage system for continuous backup of the full dedicated server. Backups are transferred from the server to the NAS at defined intervals via FTP. The client can restore full server data from the NAS if an incident occurs. The NAS is configured with RAID for redundancy so the risk of data loss is minimized. The NAS is reachable both on the local network and over the internet via static IP for remote access and management. Result: reliable, scheduled full-server backups, high assurance of data recovery, and reduced risk of data loss through RAID.

Implemented a monitoring and activity-logging system across the organization’s endpoints to observe user activity, compute personnel performance KPIs (including time gaps and utilization), and monitor changes to critical and sensitive data and accounts. The system provides live visibility into connected endpoints, calculates per-personnel performance and idle/active time, and produces monthly reports for management. Deployed both on-premises and in the cloud. Result: consistent KPI measurement for personnel, oversight of sensitive information and account changes, and a single platform for productivity and security monitoring across the organization.

Designed and implemented a web-based platform for Paladium to connect athletes and coaches. Athletes access assigned training programs, workout plans, and diet plans; they submit training logs for coach review and receive approval through the platform. The system was delivered at the club’s request during the COVID-19 pandemic. Result: the club maintained operations and revenue during lockdowns by moving training delivery and follow-up online and preserving the athlete–coach workflow remotely.

Implemented a user behavior monitoring system (session replay, heatmaps, and related analytics) to assess UI/UX quality and improve the customer journey. The system is built via integration with third-party analytics/replay tools and custom scripts that capture and process behavior data. In addition to guiding interface and flow improvements, it surfaces potential errors and exceptions along the user path so technical and product teams can fix issues. Result: data-driven UI/UX decisions, better journey optimization, and earlier detection of errors affecting users.

Designed a monitoring system that tracks file changes and suspicious activity on servers to help prevent backdoor installation and internal abuse. It connects to the Alpha servers (central monitoring servers), runs in the background, logs file and activity changes, and flags anomalies so admins can react. Result: better visibility into changes and malicious behavior and a stronger defense against backdoors and unauthorized access.

Designed and implemented a custom firewall script that analyzes user behavior to detect bots and malicious traffic and integrates with Cloudflare Turnstile to challenge or block only suspicious flows after detection. Unlike Cloudflare’s own challenge flow (which in DDoS or crisis mode can challenge all visitors), this system uses behavior-based logic so legitimate traffic—including useful crawlers such as search engines, CDN crawlers, AI crawlers, and retargeting campaigns—is not sent to Turnstile. Malicious bots and scrapers are identified and then directed to Turnstile or blocked. Result: roughly 90% reduction in bad bot traffic, freed server resources, and effective mitigation of scraping while keeping useful crawlers and real users unaffected.

Implemented a cache-warming (train-cache) system so the site is preloaded before guest traffic: the full content set is queued and reloaded so important pages and data are written into cache layers (and, where used, into Redis in RAM) before users request them. Execution is done via custom scripts and PowerShell so warming runs on a schedule or on demand. The approach is proactive—cache is filled in advance—rather than relying only on cache that fills on first visit. Result: improved performance and faster response for all users, including first hits, with warm cache and reduced cold-cache delay.

Implemented a streaming and video platform for AXOR with membership tiers, monetization (subscriptions, pay-per-view, ads), and wallet. Video upload, encoding, adaptive streaming, channels, playlists, and admin tools. Content hosted on internal/self-hosted servers. Outcome: self-hosted delivery, improved speed, and better UX.

Implemented a customized crypto gateway for the platform: per-transaction address generation via Coinbase and Gemini (and optional nodes), instant fiat/crypto conversion and transfers, multi-chain and Lightning support. Connection to the platform and custom scripts via webhooks and REST API. Security: address whitelist, AES-256, encrypted config. Outcome: one gateway for the gaming tournament platform with regenerable addresses and stable webhook integration with the platform and scripts.

Implemented a full-featured search system with content ingestion into a distributed index (configurable analyzers, tokenization, language handling), relevance tuning (scoring, boosting, faceted filters), and query understanding (synonyms, typo tolerance, optional NLP). Supports per-tenant indices and custom ranking. Result: fast, scalable search with deep indexing and tunable relevance for complex, high-traffic use cases.

Designed and implemented a layer so all third-party systems (analytics, pixels, tags, etc.) send data without blocking or delaying page load. Scripts are either hosted locally or loaded in a controlled way (deferred/async, with local caching and longer cache lifetime than default third-party domains), so the browser does not wait on external roundtrips for initial render. The same approach is applied across multiple third-party services, so many integrations can be added without cumulative slowdown. Result: pages load without extra delay from third-party calls, and a large number of third-party systems can be connected without degrading performance.

Designed and implemented a website for a real estate consultancy that operates without manual content production. The site is integrated with the MLS via API so listings in the company’s coverage area are ingested and published automatically. Listings are updated in real time with no editorial effort. Result: elimination of in-house content production, high listing volume at any time, and a strong increase in user engagement.

Implemented database sharding—splitting one large database into smaller shards, each holding a subset of data—for a high-traffic e‑commerce site (~5k products) and heavy analytical workloads (e.g. behavioral analysis, regional sales forecasting). Outcomes: horizontal scalability (add shards instead of upgrading a single server); faster queries and lighter indexes per shard; lower CPU/RAM load and fewer bottlenecks; reduced single point of failure; parallel processing across shards for analytics and reporting; less table locking under high write volume; higher write throughput for event and log-heavy systems; and more cost-effective scaling (e.g. multiple smaller servers instead of one very large one). Result: the site and analytical systems scale with traffic and data growth while keeping performance and cost under control.

Designed and implemented a multi-layer caching and serving stack for heavy, high-traffic websites: page optimization, LiteSpeed as the main server on 80/443 (SSL, HTTP/2, PHP/WordPress), Nginx on 8080 for static files and custom routing and on 8081 for CDN-style distribution, and KeyDB on 6379 for WordPress object cache in RAM. Traffic flows to LiteSpeed; static assets are offloaded to Nginx; application cache is centralized in KeyDB with no port conflicts and SSL termination only at LiteSpeed. Result: sites delivered with high performance under load, efficient use of RAM and disk, and clear separation of roles (app server, static/CDN, cache).

Built a revenue-to-cost calculation model for each campaign type, with tracking of results and conversion rate per campaign. The system flags low-return, high-cost campaigns and supports A/B testing to scale what works. E‑commerce and payment gateway were connected to analytics and conversion tracking for end-to-end measurement. Result: more efficient ad spend, identification of unproductive campaigns, and stable, revenue-positive campaigns through data-driven optimization.

Built an automation layer that updates order status for customers and notifies them via SMS, email, and the user panel. Status changes are driven by API links to shipping carriers and warehouse groups, so updates happen automatically as orders move. Result: fewer operators needed to manage and communicate order status.

Connected the online store to a CRM so the sales team and operators could manage customers faster and more accurately. High order volume makes customer management in the CMS impractical; the CRM offers stronger tools plus inventory and accounting. Result: customer and order handling moved into the CRM, with better visibility and control for sales and ops.

Built a separate script for a site with ~10,000 products that applies price and discount changes without editing products in the CMS (where updates were slow). Changes are applied directly to the main database for fast bulk updates, with bulk-edit commands. The system was extended to automatically renew or rotate discounts on a schedule 24/7. Result: the need for an operator to change or renew discounts around the clock was fully removed; constant, automated promotional urgency drove a 48% increase in sales.

Implemented an intelligent layer on the site: real-time analysis of session events, mapping of user journey to specific products, and AI-driven conversation and need assessment. The system reacts in real time with guided dialogue and product suggestions. Result: user engagement, conversion rate, and time on site increased by up to 30%; thanks to better guidance and recommendations, sales and customer conversion increased by up to 40%.

Designed and delivered a dedicated payment script that sits between the store and bank gateways: one API receives checkout data from the site, routes it to the script, and all gateways run through it. Full transaction data is stored and managed in the script. The site runs behind CDN (IP varies by user), while banks need a fixed IP—solved by hosting the payment script on a dedicated subdomain with stable IP, low-latency connection, and no change to checkout flow. Result: reliable payments, centralized transaction management, and no CDN vs. gateway conflicts.

Designed and built a flowchart-driven chat system with an AI agent for 24/7 customer support in any language. Features: real-time translation of conversations between agents and customers; online queue that activates when concurrent users rise and agent capacity is limited—estimates wait time and connects users to an agent at the right moment; AI trained on flowcharts for long, structured conversations and guiding customers toward purchase. Outcome: scalable 24/7 support, controlled wait times under load, and AI-assisted sales conversations.

Implemented a download protection layer for digital product stores: each customer receives a unique, protected download link tied to their purchase. Link generation and encryption hide the underlying URL structure and file source so links cannot be reverse‑engineered or shared to other users. Access is validated per user and per transaction. Result: controlled, non-transferable access to digital assets and reduced link sharing and piracy, so the sales process and revenue are protected.