Deen Haji's Blog

Author: deen

  • Spawn World: Revolutionizing AI-Driven Gameplay with InWorld AI

    Introduction to Spawn World

    Spawn World is an ambitious PvPvE game built on Unreal Engine 5, blending arena combat, extraction gameplay, and MMO elements into a dynamic, skill-based battlefield. The game immerses players in a living, breathing world, where battles range from adrenaline-fueled guerrilla skirmishes to large-scale warfare. But what truly sets Spawn World apart is its use of InWorld AI to power NPCs, crafting procedural, lifelike characters that evolve and react dynamically.

    https://pitch.com/v/spawn-world-x-inworld-ai-rrzeeb

    This article explores how InWorld AI transforms NPCs from static entities into living, thinking, and evolving characters—creating an immersive sandbox experience.

    The Power of Procedural NPCs

    In most games, NPCs follow predefined scripts, creating predictable and repetitive interactions. Spawn World breaks this mold by procedurally generating unique NPCs based on ten foundational attributes:

    1. Lore – A backstory that defines the character’s past.
    2. Personality – Different tones and attitudes shape interactions.
    3. Ambitions & Goals – Long-term dreams and short-term objectives drive NPC behavior.
    4. Affinity – Biases influence decisions and relationships.
    5. World Knowledge – Understanding of the world that grows over time.
    6. Lineage – Traits inherited from ancestors.
    7. Society – NPCs form communities and civilizations.
    8. Evolution – Characters change based on experiences.
    9. Skills & Equipment – Unique abilities and gear based on the world.
    10. Appearance – Physical features influenced by genetics and society.

    These factors work together to create diverse NPCs that don’t just exist—they live.

    NPC Lore: A Story That Writes Itself

    NPCs in Spawn World are not generic background characters; each has a procedurally generated life story.

    For example, consider a blue orc born in a remote village. He starts as a simple farmer but dreams of becoming a knight. His journey takes him to a regiment where he proves his worth in combat, rising through the ranks. Eventually, he finds himself in a life-or-death duel with the player. If he wins, his reputation soars, earning him a place as the king’s right-hand guard. However, as he serves, he learns of the king’s cruelty—leading to a fateful moment of betrayal where he seizes the throne.

    This story isn’t pre-written—it emerges dynamically based on AI-driven choices.

    NPC Personality: Conversations That Feel Real

    InWorld AI enables unique conversational experiences by giving NPCs distinct personalities.

    • A stern knight may speak in a commanding tone.
    • A carefree mercenary might joke and taunt during battle.
    • Body language and animations reinforce their personality—a seasoned warrior walks with a confident swagger, while a rookie fumbles nervously.

    Every NPC feels authentic and reactive, ensuring no two interactions are ever the same.

    Long-Term Ambitions & Short-Term Goals

    NPCs in Spawn World are driven by ambition—a deep-rooted motivation shaping their destiny.

    • Long-term ambitions define where an NPC wants to go in life.
    • Short-term goals dictate how they get there—whether through battle, diplomacy, or trade.

    For example, an NPC might aspire to be a legendary blacksmith. But first, they need to learn metalworking, acquire rare ores, and forge their first masterpiece. Players might help or hinder their journey, influencing their fate and relationship with the world.

    Affinity: The Emotional AI Factor

    NPCs develop personal biases based on their experiences and ancestry.

    • Gained Affinity – An NPC might develop a love for swords after training with a warrior.
    • Inherited Affinity – A character may be naturally inclined toward magic due to their lineage.

    These emotional drivers shape the choices NPCs make, ensuring no two AI-driven characters react the same way.

    World Knowledge: Expanding the Horizon

    Characters learn and adapt based on the world around them.

    • Knowledge sets limits on what they believe is possible.
    • As NPCs learn new things, they evolve their strategies—an inexperienced bandit may initially attack recklessly but later become cunning and tactical.

    This ensures that the AI-driven world never stagnates—it is always growing, changing, and adapting.

    Lineage: The Power of Generations

    NPC traits pass down through generations, creating a dynamic bloodline system.

    • A warrior’s child inherits their fighting instinct.
    • A scholar’s descendant might excel in magic.
    • World knowledge evolves, leading to societal growth over generations.

    This means players can interact with NPCs over different lifetimes, seeing the impact of past choices ripple through history.

    NPC Societies: Living Civilizations

    NPCs don’t just exist in isolation—they build settlements, trade, and form political alliances.

    • Settlements evolve as NPCs gather resources and expand.
    • Trade networks emerge based on supply and demand.
    • Politics plays a role—NPC factions ally, betray, or declare war.

    This creates a true simulation where the world continues to change whether the player is involved or not.

    NPC Evolution: A World That Grows

    Spawn World’s NPCs evolve on three levels:

    1. Individual Evolution – Characters grow based on their experiences.
    2. Generational Evolution – Future NPCs inherit skills, knowledge, and ambitions.
    3. Societal Evolution – As civilizations expand, new technologies, cultures, and conflicts arise.

    This ensures that every session in Spawn World feels fresh, as AI-driven changes unfold naturally.

    Skills & Equipment: Shaped by the World

    NPCs in Spawn World don’t just spawn with random gear—their equipment is influenced by local resources, technology, and trade.

    • A nomadic tribe may favor bows and lightweight armor.
    • A kingdom rich in iron will produce heavily armored knights.
    • A magical enclave might focus on staffs and enchanted relics.

    This dynamic system ensures NPCs feel grounded in their environment, rather than being generic spawns.

    Appearance: More Than Just Skins

    NPC appearances are genetically influenced by their lineage.

    • Facial structures, skin tones, and body types vary based on ancestry.
    • Fashion and armor reflect cultural and societal influences.

    No two NPCs will ever look exactly alike, adding to the richness of the world.

    Conclusion: AI-Driven Immersion Like Never Before

    Spawn World is more than just a game—it’s a next-generation AI-driven simulation where every NPC is a dynamic character with its own life, ambitions, and evolution.

    With InWorld AI powering its NPCs, players can experience truly immersive interactions, where no two playthroughs are ever the same. This isn’t just about fighting enemies—it’s about living in a world that feels real, reacts dynamically, and grows over time.

    Welcome to Spawn World. Your story is just beginning.

  • Scalable Multiplayer Game Infrastructure with Node Mesh Networking: Putting the ‘massive’ back into MMOs

    Building a large-scale multiplayer game requires an infrastructure capable of handling thousands of concurrent players while maintaining a seamless experience. This article outlines a robust, horizontally scalable server architecture using a node mesh network, TCP-based inter-server communication, and dynamic zone management to distribute player load efficiently.

    Node Mesh Network and TCP Communication

    At the core of this infrastructure is a node mesh network of interconnected servers communicating over TCP. This enables a controller-worker model, where a central controller node manages worker nodes responsible for handling gameplay within specific map regions.

    To ensure high availability, the system employs a round-robin election mechanism—if the controller node fails, a worker node can be dynamically elected to take over its role.

    Zone-Based Load Balancing and State Synchronization

    The game world is divided into zones, each managed by a specific server. The number of zones can be adjusted based on the available servers, allowing for flexible load balancing. High-traffic areas can be subdivided into smaller zones, dynamically distributing server load.

    Each server synchronizes its state with neighboring nodes, enabling entity replication across zones. This allows players to see entities beyond their current server boundary, ensuring a seamless multiplayer experience.

    Entity Authority and Cross-Server Interaction

    Entity ownership is strictly managed to maintain server authority:

    • Each entity is owned by a specific server, which has full write access.
    • Other servers only store read-only replicas of the entity.
    • When a player interacts with an entity in another server’s zone (e.g., attacking an enemy), the local server relays the action to the authoritative server, ensuring true server-side validation for events like hit detection and damage calculations.

    This design prevents cheating and ensures that no single client can override game logic.

    Horizontal Scalability and Real-Time Load Balancing

    A key advantage of this infrastructure is its horizontal scalability:

    1. Adding more servers: New nodes can be added to manage more zones or subdivide crowded regions dynamically.
    2. Run-time server allocation: Zones can be reallocated in real-time based on player distribution, improving performance and reducing latency.
    3. Optimized geo-location: Servers can be placed closer to high-population areas, reducing response times and enhancing gameplay.

    Database Replication, Monitoring, and Orchestration

    To support real-time persistence, each node runs PostgreSQL database replication, ensuring low-latency database access without overwhelming a central server.

    For monitoring and orchestration:

    • Prometheus is used for real-time server resource monitoring, tracking CPU, memory, and network usage.
    • Kubernetes manages the cluster, ensuring automatic recovery, scaling, and failover handling.
    • Docker provides containerization for deploying the game’s backend infrastructure efficiently.

    Technology Stack

    • Node Server: Golang, ensuring fast, concurrent processing.
    • Unreal Engine Server Connector: C++, using memory-mapped files for inter-process communication (IPC) between the game server and node server.

    Conclusion

    By leveraging a node mesh architecture, dynamic zone-based load balancing, and a robust inter-server communication model, this infrastructure provides high scalability, fault tolerance, and optimal performance. This design ensures that as player numbers grow, the system can dynamically expand, providing a smooth and responsive experience across a distributed game world.

  • Automated Marketing: Generating gaming traffic with giveaways

    In 2016 I created a game giveaway community called JustLoot.Me. It began from an idea to build up a gaming audience to promote my own game I had conceptualised a few days earlier (Spawn.world). I started a Steam group and found some giveaway sites which required steam signups to enter giveaways. I found that they display the steam username of everybody who enters these giveaways. Since I knew these users were interested in game giveaways, I came up with a strategy to start inviting all of them to my own steam group through my own automation I created. I ended up getting 200 users on the first day and consistently an additional 200 on each following day. I was giving away popular games (e.g Dark Souls) via a platform called Gleam which allows me to require joining my steam group in order to enter the giveaway. There was also a referral feature which gave them extra chances to win the more users they invited to enter. After implementing this, my growth was exponential and I was growing by 500 users each day instead of just the 200 from the automation. I reached around 20k users within a month. I still have the email list to this day (Gleam allowed me to collect their email) – planning to use it to promote my own game Spawn. I’ll be relaunching JustLoot.Me as a platform to create viral giveaways which other giveaway communities and streamers can utilise very soon.

  • Automated Marketing: Building a recommendation engine for influencers/brands

    In 2015, I created automation to grow my instagram account. I quickly gained 5k followers and reached 1500 likes per picture. It made me realise the power of social media and the ability to generate a lot of targeted traffic which can be utilised by brands. I researched influencer marketing and devised a strategy to match brands and influencers together based on niche. In order to do this, I had to collect a lot of data. I create software which extracted instagram profile data including the bio, tags published, follower count, average likes, average comments and even the tags used by the engagers of the profile to determine the interests of their actual audiences. Using this data, I was able to create SQL queries to get targeted lists of users within any niche. This was great, but not optimal – I want to go even further and started researching data analysis and machine learning. I scored the weight of tags used by the ratio of times they were mentioned within an individual’s profile – now for each profile I had a combination of tags and ratios. Using this data with algorithms such as KNN and ALS I was able to create an accurate ML model to infer the niche of any profile in real-time. I created an online ML solution using Apache Spark allowing high throughput and a constantly evolving model via a feedback loop. I ended up developing a SaaS solution to match brands and influencers based on their niche using this model, which was used for highly targeted influencer campaigns leading to better conversions/ROIs.

  • How I started with automation and AI

    In 2011 after I left high school at 16, I started learning about marketing so I could 1. Earn money remotely online and 2. Use this knowledge to market my own businesses I planned to start. After learning the basics of SEO, I realised that backlinks were staple in ranking to the top of Google and also that the process of creating these backlinks was similar across all websites within certain platforms (e.g Angel forums). I devised a plan to automate this and sell it as a service, so I researched and came across a software called Zennoposter which allows you to create automation. I quickly learned how to use Zennoposter, staying up all day and night for a few weeks straight, working away at trying to automate this strategy I came up with and eventually had a working product. I began selling a backlink service on Fiverr, providing 500 backlinks for $5, or 2000 dripfed over X days for $20. Orders started flooding in and I realised I was spending a lot of time generating reports for each client, so I automated the process of getting the client’s website data from each Fiverr order to create the backlinks and the process of delivering reports each day. Now I had built a fully automated passive income stream.